ChatGPT is designed to be useful, trustworthy, and adaptable—so you can make it your own.

OpenAI joins the EU Code of Practice, advancing responsible AI while partnering with European governments to drive innovation, infrastructure, and economic growth.

Building a family of AI products for everyone.

OpenAI partners with the American Federation of Teachers to launch a 5-year initiative equipping 400,000 K-12 educators to lead AI innovation in classrooms.

Learn how Genspark built a $36M ARR AI product in 45 days—with no-code agents powered by GPT-4.1 and OpenAI Realtime API.

Today, OpenAI, in partnership with Mandala Partners, is sharing the OpenAI AI Economic Blueprint for Australia. At a time when boosting productivity has emerged as a national priority for Australia, the Blueprint provides a clear, actionable plan for how Australia can unlock the full economic and social potential of artificial intelligence.

Retell AI is transforming the call center with AI voice automation powered by GPT-4o and GPT-4.1. Its no-code platform enables businesses to launch natural, real-time voice agents that cut call costs, boost CSAT, and automate customer conversations—without scripts or hold times.

Unify, an AI-powered GTM platform, uses OpenAI’s o3, GPT-4.1, and CUA to automate prospecting, research, and outreach. With hyper-personalized messaging and an always-on workflow, Unify helps teams generate pipeline at scale while focusing on high-impact customer interactions.

We study how training on incorrect responses can cause broader misalignment in language models and identify an internal feature driving this behavior—one that can be reversed with minimal fine-tuning.

Advanced AI can transform biology and medicine—but also raises biosecurity risks. We’re proactively assessing capabilities and implementing safeguards to prevent misuse.

We’re launching OpenAI for Government, a new initiative focused on bringing our most advanced AI tools to public servants across the United States. We're supporting the U.S. government's efforts in adopting best-in-class technology and deploying these tools in service of the public good.

OpenAI and Mattel are partnering to integrate AI into iconic brands such as Barbie and Hot Wheels, aiming to enhance creative development, streamline workflows, and create new ways for fans to engage.

OpenAI introduces its Outbound Coordinated Disclosure Policy to guide how it responsibly reports vulnerabilities in third-party software—emphasizing integrity, collaboration, and proactive security at scale.

Outbound coordinated vulnerability disclosure policy

OpenAI is fighting a court order at the demands of The New York Times and plaintiffs, which involves retention of consumer ChatGPT and API user data indefinitely. Learn how we’re working to uphold user privacy, address legal requirements, and stay true to our data protection commitments.

In our June 2025 update, we outline how we’re disrupting malicious uses of AI—through safety tools that detect and counter abuse, support democratic values, and promote responsible AI deployment for the benefit of all.

Wix’s AI Website Builder, powered by OpenAI, lets anyone create a full website in minutes—just by describing their idea in a conversation.

We are replacing the existing GPT-4o-based model for Operator with a version based on OpenAI o3. The API version will remain based on 4o.

OpenAI announces the opening of its first office in Germany, based in Munich.

CodeRabbit uses OpenAI models to revolutionize code reviews—boosting accuracy, accelerating PR merges, and helping developers ship faster with fewer bugs and higher ROI.

We’re launching Stargate UAE – the first international deployment of Stargate, OpenAI’s AI infrastructure platform.

New features in the Responses API: Remote MCP, image gen, Code Interpreter, and more. Powering faster, smarter agents with GPT-4o & o-series models, plus new features for reliability and efficiency.

Codex is a cloud-based coding agent. Codex is powered by codex-1, a version of OpenAI o3 optimized for software engineering. codex-1 was trained using reinforcement learning on real-world coding tasks in a variety of environments to generate code that closely mirrors human style and PR preferences, adheres precisely to instructions, and iteratively runs tests until passing results are achieved.

Introducing Codex: a cloud-based software engineering agent that can work on many tasks in parallel, powered by codex-1. With Codex, developers can simultaneously deploy multiple agents to independently handle coding tasks such as writing features, answering questions about your codebase, fixing bugs, and proposing pull requests for review.

A conversation with Jochen Koedijk, Chief Marketing Officer of Expedia Group.

HealthBench is a new evaluation benchmark for AI in healthcare which evaluates models in realistic scenarios. Built with input from 250+ physicians, it aims to provide a shared standard for model performance and safety in health.

Read the message Sam shared with the company earlier today.

Why infrastructure is destiny and how the US can seize it.

Data residency builds on OpenAI’s enterprise-grade data privacy, security, and compliance programs supporting customers worldwide.

Discover how the San Antonio Spurs are using custom GPTs to enhance fan engagement, streamline operations, and drive innovation across teams.

Lowe’s partnered with OpenAI to build Mylow and Mylow Companion, AI-powered tools that bring expert help to both customers and store associates—making complex home improvement projects easier to plan, navigate, and complete.

A new initiative to support countries around the world that want to build on democratic AI rails.

Sam Altman has written that we are entering the Intelligence Age, a time when AI will help people become dramatically more capable. The biggest problems of today—across science, medicine, education, national defense—will no longer seem intractable, but will in fact be solvable. New horizons of possibility and prosperity will open up.

John Deere’s Justin Rose talks about transforming agriculture with AI and shares how the company is scaling innovation to help farmers work smarter, more efficiently, and sustainably.

An update from the OpenAI board on transitioning its for-profit entity to a Public Benefit Corporation, reinforcing its mission-driven structure under nonprofit oversight while enabling greater impact and long-term alignment with the public good.

A conversation with Chandhu Nair, Senior Vice President of Data, AI, and Innovation.

A deeper dive on our findings, what went wrong, and future changes we’re making.

We have rolled back last week’s GPT‑4o update in ChatGPT so people are now using an earlier version with more balanced behavior. The update we removed was overly flattering or agreeable—often described as sycophantic.

Watch hands-on demos of the lastest in ChatGPT for Business: o3, image generation, enhanced memory, and internal knowledge.

Our latest image generation model is now available in the API via ‘gpt-image-1’—enabling developers and businesses to build professional-grade, customizable visuals directly into their own tools and platforms.

A conversation with Connor Zwick, CEO & Co-founder of Speak.

The Washington Post is partnering with with OpenAI to integrate news into ChatGPT, providing users with summaries, quotes, and direct links to original reporting.

Our smartest and most capable models to date with full tool access

OpenAI o3 and o4-mini represent a significant breakthrough in visual perception by reasoning with images in their chain of thought.

OpenAI o3 and OpenAI o4-mini combine state-of-the-art reasoning with full tool capabilities—web browsing, Python, image and file analysis, image generation, canvas, automations, file search, and memory.

OpenAI is appointing four new advisors to help inform OpenAI’s philanthropic efforts.

Sharing our updated framework for measuring and protecting against severe harm from frontier AI capabilities.

Introducing GPT-4.1 in the API—a new family of models with across-the-board improvements, including major gains in coding, instruction following, and long-context understanding. We’re also releasing our first nano model. Available to developers worldwide starting today.

BrowseComp: a benchmark for browsing agents.

Advancing model performance and real world evaluation in applied domains.

A conversation with Cameron Adams, Chief Product Officer and Co-founder of Canva.

Today, OpenAI is sharing the EU Economic Blueprint—a set of proposals to help Europe seize the promise of artificial intelligence, drive sustainable economic growth across the region, and ensure that AI is developed and deployed by Europe, in Europe, for Europe.

Notion is a connected workspace where teams write, plan, and organize everything from meeting notes to product roadmaps. Today, it’s also a deeply AI-powered platform, used by millions to summarize content, generate writing, and ask questions in natural language across their entire workspace.

Already a nonprofit, and already using AI to help people solve hard problems, OpenAI aims to build the best-equipped nonprofit the world has ever seen—combining potentially historic financial resources with something even more powerful: technology that can scale human ingenuity itself.

We introduce PaperBench, a benchmark evaluating the ability of AI agents to replicate state-of-the-art AI research.

Recommendations for pro-innovation policies that can help make the UK the AI capital of Europe.

Today we’re announcing new funding—$40B at a $300B post-money valuation, which enables us to push the frontiers of AI research even further, scale our compute infrastructure, and deliver increasingly powerful tools for the 500 million people who use ChatGPT every week.

Moving from intent-based bots to proactive AI agents.

At OpenAI, we proactively adapt, including by building comprehensive security measures directly into our infrastructure and models.

At OpenAI, we have long believed image generation should be a primary capability of our language models. That’s why we’ve built our most advanced image generator yet into GPT‑4o. The result—image generation that is not only beautiful, but useful.

4o image generation is a new, significantly more capable image generation approach than our earlier DALL·E 3 series of models. It can create photorealistic output. It can take images as inputs and transform them.

Hebbia’s deep research automates 90% of finance and legal work, powered by OpenAI

Online resource hub will support AI literacy and help people from all backgrounds access tools, best practices, and peer insights to use AI.

OpenAI has grown a lot. We remain focused on the same core—pursuing frontier AI research that accelerates human progress–but we now also deliver products used by hundreds of millions of people.

An OpenAI and MIT Media Lab Research collaboration.

By integrating its data systems with OpenAI’s LLMs, Booking.com delivers smarter search, faster support, and intent-driven travel experiences.

For the first time, developers can also instruct the text-to-speech model to speak in a specific way—for example, “talk like a sympathetic customer service agent”—unlocking a new level of customization for voice agents.

A conversation with Minna Song, CEO & Co-founder of EliseAI.

Join us as we share our latest releases and how ChatGPT is becoming more interactive, customized to the way your teams work, and agentic.

We welcome the court’s March 4, 2025, decision rejecting Elon Musk’s latest attempt to slow down OpenAI for his personal benefit.

Recommendations build on OpenAI’s Economic Blueprint to strengthen America’s AI leadership.

LY Corporation: Driving growth and ‘WOW’ moments with OpenAI

We’re evolving our platform to help developers and enterprises build useful and reliable agents.

Frontier reasoning models exploit loopholes when given the chance. We show we can detect exploits using an LLM to monitor their chains-of-thought. Penalizing their “bad thoughts” doesn’t stop the majority of misbehavior—it makes them hide their intent.

Nubank elevates customer experiences with OpenAI

Accelerating engineering cycles 20% with OpenAI.

A conversation with Claire Vo, Chief Product Officer of LaunchDarkly, about the changing role of product managers, her anti-to-do list, and building AI-native teams.

OpenAI commits $50M in funding and tools to leading institutions.

OpenAI and nine national labs bring together leading scientists for first-of-its kind event.

Mercari leverages GPT-4o mini and GPT-4 to streamline selling, enhance product listings, and boost sales, transforming the online marketplace with features like AI Listing Support and Mercari AI Assistant.

We’re releasing a research preview of OpenAI GPT‑4.5, our largest and most knowledgeable model yet.

We’re releasing a research preview of GPT‑4.5—our largest and best model for chat yet. GPT‑4.5 is a step forward in scaling up pre-training and post-training.

Endex builds the future of financial analysis, powered by OpenAI’s reasoning models.

This report outlines the safety work carried out prior to releasing deep research including external red teaming, frontier risk evaluations according to our Preparedness Framework, and an overview of the mitigations we built in to address key risk areas.

Estonia and OpenAI to bring ChatGPT to schools nationwide. OpenAI will work with the Estonian Government to provide students and teachers in the secondary school system with access to ChatGPT Edu.

Ensuring AI benefits humanity by advancing democratic AI, preventing misuse, and protecting against authoritarian threats.

A conversation with Jai Malkani, Head of AI and Product, Customer Obsession at Uber.

A look into state-by-state adoption and how gaps might impact workforce readiness.

Can frontier LLMs earn $1 million from real-world freelance software engineering?

OpenAI and Guardian Media Group announce content partnership to bring Guardian news content to ChatGPT.

A conversation with Andrea Ellis, Chief Financial Officer of Fanatics Betting and Gaming.

A conversation with Fiona Tan, Chief Technology Officer of Wayfair.

Rogo scales AI-driven financial research with OpenAI o1

We’ve made updates to the Model Spec based on external feedback and our continued research in shaping desired model behavior.

OpenAI and Schibsted Media Group announce content partnership to bring Guardian news and archive content to ChatGPT.

We aired our first-ever television ad during the Super Bowl to pique people’s curiosity and help us all realize how AI can open up new possibilities for us, create more fulfillment in our lives, and make us more productive, just as all the tools that came before AI did for those who came before us.

OpenAI looks forward to engaging with global leaders on AI’s role in shaping innovation and economic prosperity.

Data residency builds on OpenAI’s enterprise-grade data privacy, security, and compliance programs supporting customers worldwide.

The largest deployment of ChatGPT to date will expand the use of AI in education and help the United States build an AI-ready workforce.

Using ChatGPT to catch halibut

Using ChatGPT to find inspiration for nail art

ChatGPT and personal tutoring

An agent that uses reasoning to synthesize large amounts of online information and complete multi-step research tasks for you. Available to Pro users today, Plus and Team next.

How OpenAI deep research helps Bain & Company understand complex industry trends.

Pushing the frontier of cost-effective reasoning.

This report outlines the safety work carried out for the OpenAI o3-mini model, including safety evaluations, external red teaming, and Preparedness Framework evaluations.

OpenAI’s latest line of reasoning models will be used by nation’s leading scientists to drive scientific breakthroughs.

ChatGPT Gov is designed to streamline government agencies’ access to OpenAI’s frontier models.

Drawing from OpenAI’s established safety frameworks, this document highlights our multi-layered approach, including model and product mitigations we’ve implemented to protect against prompt engineering and jailbreaks, protect privacy and security, as well as details our external red teaming efforts, safety evaluations, and ongoing work to further refine these safeguards.

A universal interface for AI to interact with the digital world.

A research preview of an agent that can use its own browser to perform tasks for you. Available to Pro users in the U.S.

Bertelsmann, the global media, services, and education company headquartered in Germany, will integrate OpenAI’s technology across multiple brands around the world.

Trading Inference-Time Compute for Adversarial Robustness

Announcing The Stargate Project

OpenAI, and our strategic partners, are thrilled about our shared vision for the Infrastructure of AGI. We are energized by the challenges we face and are excited by the prospect of partnering with firms across the industrial base to deliver against our ambitious mission. Specifically, we want to connect with firms across the built data center infrastructure landscape, from power and land to construction to equipment, and everything in between.

The power of personalized AI

Publishers representing hundreds of newsrooms and content brands are using OpenAI partnerships and grant programs to adopt AI tools and strengthen the news ecosystem, while ChatGPT users gain access to information from leading, reliable publications.

Adebayo Ogunlesi Joins OpenAI’s Board of Directors

OpenAI’s Economic Blueprint

A stronger non-profit supported by the for-profit’s success.

Deliberative alignment: reasoning enables safer language modelsIntroducing our new alignment strategy for o1 models, which are directly taught safety specifications and how to reason over them.

Introducing OpenAI o1, Realtime API improvements, a new fine-tuning method and more for developers.

Elon Musk’s latest legal filing against OpenAI marks his fourth attempt in less than a year to reframe his claims. However, his own words and actions speak for themselves—in 2017, Elon not only wanted, but actually created, a for-profit as OpenAI’s proposed new structure.

Zalando boosts the customer experience with its Assistant, powered by GPT-4o mini

Our video generation model, Sora, is now available to use at sora.com. Users can generate videos up to 1080p resolution, up to 20 sec long, and in widescreen, vertical or square aspect ratios. You can bring your own assets to extend, remix, and blend, or generate entirely new content from text.

Filmmaker Lyndon Barrois describes how to use Sora as a storytelling tool.

Interdisciplinary artist Minne Atairu discusses how Sora helps realize her vision.

Filmmaking duo Vallée Duhamel explains how Sora helps build new worlds.

Sora is OpenAI’s video generation model, designed to take text, image, and video inputs and generate a new video as an output. Sora builds on learnings from DALL-E and GPT models, and is designed to give people expanded tools for storytelling and creative expression.

Put AI to work for your product team

Broadening usage of frontier AI

This report outlines the safety work carried out prior to releasing OpenAI o1 and o1-mini, including external red teaming and frontier risk evaluations according to our Preparedness Framework.

OpenAI and Future, the global platform for specialist media, have today announced a strategic partnership to bring content from Future’s 200 plus media brands to OpenAI’s users.

Morgan Stanley uses AI evals to shape the future of financial services

Advancing red teaming with people and AI

Empowering a global org with ChatGPT

Building smarter maps with GPT-4o vision fine-tuning

By combining commercial experience and deep LLM expertise with OpenAI’s models, Rox makes every seller a top 1% seller.

Our first office in continental Europe

A Student’s Guide to Writing with ChatGPT

Data-driven beauty: How The Estée Lauder Companies unlocks insights with ChatGPT

This comment was submitted in response to a request for information from the National Telecommunications and Information Administration (NTIA).

Get fast, timely answers with links to relevant web sources

Promega's top-down adoption of ChatGPT accelerates manufacturing, sales, and marketing

Put AI to Work for Marketing Teams

A factuality benchmark called SimpleQA that measures the ability for language models to answer short, fact-seeking questions.

Decagon and OpenAI deliver high-performance, fully automated customer support at scale

OpenAI’s approach to AI and national security

We’ve simplified, stabilized, and scaled continuous-time consistency models, achieving comparable sample quality to leading diffusion models, while using only two sampling steps.

OpenAI appoints Scott Schools as Chief Compliance Officer

Dr. Ronnie Chatterji named OpenAI’s first Chief Economist

OpenAI and the Lenfest Institute AI Collaborative and Fellowship program

In this video, we share how the o1 reasoning models can help in domains like coding, strategy, and research.

We've analyzed how ChatGPT responds to users based on their name, using AI research assistants to protect privacy.

We introduce MLE-bench, a benchmark for measuring how well AI agents perform at machine learning engineering.

OpenAI’s mission is to ensure that artificial general intelligence benefits all of humanity. We are dedicated to identifying, preventing, and disrupting attempts to abuse our models for harmful ends.

Hearst’s iconic brands bring curated lifestyle and local news content to OpenAI’s products.

We’re publishing our Raising Concerns Policy, which protects employees’ rights to make protected disclosures.

Introducing canvas

In addition to securing $6.6 billion in new funding from leading investors, we have established a new $4 billion credit facility with leading banks, including JPMorgan Chase, Citi, Goldman Sachs, Morgan Stanley, Santander, Wells Fargo, SMBC, UBS, and HSBC.

We are making progress on our mission to ensure that artificial general intelligence benefits all of humanity.

Developers can now build fast speech-to-speech experiences into their applications

Developers can now fine-tune GPT-4o with images and text to improve vision capabilities

Offering automatic discounts on inputs that the model has recently seen

Fine-tune a cost-efficient model with the outputs of a large frontier model–all on the OpenAI platform

Altera uses GPT-4o to build a new area of human collaboration

Put AI to work: Automate and Scale Financial Operations

We’re introducing a new model built on GPT-4o that is more accurate at detecting harmful text and images, enabling developers to build more robust moderation systems.

Minnesota’s Enterprise Translation Office uses ChatGPT to bridge language gaps

OpenAI and GEDI announce strategic partnership to bring Italian-language news content to ChatGPT.

Mercado Libre introduces Verdi, an AI developer platform powered by GPT-4o

New initiative will fuel innovation by investing in developers and organizations leveraging AI, starting in low- and middle-income countries.

Genmab embraces ChatGPT Enterprise, supported by OpenAI’s commitment to security and privacy

Improving teaching and learning in Brazil

An update on our safety & security practices

Introducing OpenAI o1

We are introducing OpenAI o1, a new large language model trained with reinforcement learning to perform complex reasoning. o1 thinks before it answers—it can produce a long internal chain of thought before responding to the user.

Advancing cost-efficient reasoning

OpenAI o1 Contributions

OpenAI o1 system card external testers acknowledgements

Quantum physicist Mario Krenn uses OpenAI o1 to help answer life's biggest questions.

Geneticist Catherine Brownstein demonstrates how OpenAI o1 can speed up the process of diagnosing rare medical challenges.

Economist Tyler Cowen explains how OpenAI o1 tackles complex economic questions.

Scott Wu, CEO and Co-Founder of Cognition, explains how OpenAI o1 makes coding decisions in a more human-like way.

Put AI to Work: Lessons from Hundreds of Successful Deployments

Ada uses GPT-4 to deliver a new customer service standard

Arizona State University embraces ChatGPT campus-wide to personalize learning, advance research, and prepare students for the future

Fine-Tuning GPT-4o Webinar

Condé Nast

Upwork puts AI to work, uniting team members, operations and product development

Fine-tune custom versions of GPT-4o to increase performance and accuracy for your applications

We banned accounts linked to a covert Iranian influence operation using ChatGPT to generate website and social media content focused on multiple topics, including the U.S. presidential campaign. We have seen no indication that this content reached a meaningful audience.

Indeed, whose mission is to help people get jobs, is the world’s #1 job site. Over 350 million unique visitors come to Indeed every month to connect with more than 3.5 million employers and over 32 million jobs. But what’s more is that every three seconds someone gets hired on Indeed.

AI can enrich lives through beauty and creativity, and its artistic potential shines in "Sleeping Beauties: Reawakening Fashion," a collaborative exhibit from The Met's Costume Institute.

We’re releasing a human-validated subset of SWE-bench that more reliably evaluates AI models’ ability to solve real-world software issues.

Zico Kolter Joins OpenAI’s Board of DirectorsWe’re strengthening our governance with expertise in AI safety and alignment. Zico will also join the Safety & Security Committee

GPT-4o system card external testers acknowledgements

In this video, we share practical examples of how employees can use ChatGPT Enterprise to efficiently analyze data and uncover insights.

This report outlines the safety work carried out prior to releasing GPT-4o including external red teaming, frontier risk evaluations according to our Preparedness Framework, and an overview of the mitigations we built in to address key risk areas.

Rakuten Pairs Data with AI to Unlock Customer Insights and Value

We are introducing Structured Outputs in the API—model outputs now reliably adhere to developer-supplied JSON Schemas.

We’re sharing a preliminary overview of the EU AI Act including upcoming deadlines and requirements, with a particular focus on prohibited and high-risk use cases

We’re testing SearchGPT, a temporary prototype of new search features that give you fast and timely answers with clear and relevant sources.

We've developed and applied a new method leveraging Rule-Based Rewards (RBRs) that aligns models to behave safely without extensive human data collection.

Introducing the most cost-efficient small model in the market

Compliance API integrations, SCIM, and GPT controls to support compliance programs, data security, and user access at scale

desc

OpenAI and Los Alamos National Laboratory are working to develop safety evaluations to assess and measure biological capabilities and risks associated with frontier models.

CriticGPT, a model based on GPT-4, writes critiques of ChatGPT responses to help human trainers spot mistakes during RLHF

We’re partnering with TIME and its 101 years of archival content to enhance responses and provide links to stories on Time.com

OpenAI Acquires Rockset

Highlighting innovative research and AI integration in cybersecurity

Diffusion models have significantly advanced the fields of image, audio, and video generation, but they depend on an iterative sampling process that causes slow generation.

We present a holistic approach to building a robust and useful natural language classification system for real-world content moderation.

Consistency models are a nascent family of generative models that can sample high quality data in one step without the need for adversarial training.

Paf adopted ChatGPT Enterprise across its entire company, with engineers using custom GPTs on a daily basis to speed up routine development tasks. Paf also integrated ChatGPT Enterprise into the grit:lab coding academy (gritlab.ax), training the next generation of software developers using an AI-augmented, systems-architecture mindset from day one. In addition to the wide range of use cases for developers and grit:lab students, 70% of Paf employees actively use ChatGPT Enterprise, spanning business teams like finance, HR, marketing, and customer support.

Color Health is working with OpenAI to pioneer a new way of accelerating cancer patients’ access to treatment. Their new Cancer Copilot application uses GPT-4o to identify missing diagnostics and create tailored workup plans, enabling healthcare providers to make evidence-based decisions about cancer screening and treatment.

Nakasone brings cybersecurity experience to growing Board of Directors; will join the Board’s Safety and Security Committee

OpenAI and Apple announce partnership to integrate ChatGPT into Apple experiences.

OpenAI welcomes Sarah Friar (CFO) and Kevin Weil (CPO)

Exploring the technology behind our text-to-speech model.

Using new techniques for scaling sparse autoencoders, we automatically identified 16 million patterns in GPT-4's computations.

We outline our architecture that supports the secure training of frontier models.

We’ve terminated accounts linked to covert influence operations; no significant audience increase due to our services.

We’re launching a new initiative to enhance the accessibility of our tools for nonprofit organizations, including discounted rates for ChatGPT Team and Enterprise.

An affordable offering for universities to responsibly bring AI to campus.

MavenAGI is a new software company for the AI era. They recently launched an AI customer service agent, built on the flexibility of GPT-4, which a number of companies like Tripadvisor, Clickup and Rho are already using to save time and better serve their customers.

We’re collaborating with WAN-IFRA, the World Association of News Publishers, to launch a global accelerator program that will assist over 100 news publishers to explore and integrate AI in their newsroom.

The Atlantic is announcing a strategic content and product partnership with OpenAI, which positions The Atlantic as a premium news source within OpenAI. The Atlantic’s articles will be discoverable within OpenAI’s products, including ChatGPT, and as a partner, The Atlantic will help to shape how news is surfaced and presented in future real-time discovery products.

In a multi-faceted agreement, Vox Media’s content will enhance the output of OpenAI’s ChatGPT, and the company will build on OpenAI’s technology to develop products to better serve its audiences and advertisers.

Companies Join Forces to Enrich OpenAI’s Generative AI Products and Platforms with Premium Journalism

Artificial general intelligence has the potential to benefit nearly every aspect of our lives—so it must be developed and deployed responsibly.

How the voices for ChatGPT were chosenWe worked with industry-leading casting and directing professionals to narrow down over 400 submissions before selecting the 5 voices.

Improvements to data analysis in ChatGPTInteract with tables and charts and add files directly from Google Drive and Microsoft OneDrive.

OpenAI and Reddit Partnership We’re bringing Reddit’s unique content to ChatGPT and our products.

Canva is a visual communication platform, enjoyed by more than 175 million people monthly to make presentations, videos, documents, websites, social media graphics and more. A majority of the world’s knowledge workers lack design training, but Canva’s combination of an easy-to-use interface, vast libraries, and time-saving tools allows anyone to create visually compelling content.

Collaborating with Carlyle to Chart the Future of Private Equity

We’re announcing GPT-4 Omni, our new flagship model which can reason across audio, vision, and text in real time.

Introducing GPT-4o and making more capabilities available for free in ChatGPT.

Introducing GPT-4o and more tools to ChatGPT free usersWe are launching our newest flagship model and making more capabilities available for free in ChatGPT.

Today we’re introducing new technology to help researchers identify content created by our tools and joining the Coalition for Content Provenance and Authenticity Steering Committee to promote industry standards.

Just over a year after launching ChatGPT, AI is changing how we live, work and learn. It’s also raised important conversations about data in the age of AI. More on our approach, a new Media Manager for creators and content owners, and where we’re headed.

API Partnership with Stack Overflow Stack Overflow and OpenAI today announced a new API partnership that will empower developers with the collective strengths of the world’s leading knowledge platform for highly technical content with the world’s most popular LLM models for AI development.

Securing advanced AI systems will require an evolution in infrastructure security. We’re calling for research and investment in six security measures that we believe will play key roles in protecting advanced AI.Protecting, exploring, and applying advanced artificial intelligence (AI) is our strategic imperative. OpenAI’s mission is to deliver positive impact of advanced AI to everything from healthcare to science to education – and yes, even to cybersecurity. That work begins with building secure, trustworthy AI systems and protecting the underlying technologies from those who seek to subvert our work to cause harm.

We will also collaborate on new AI experiences for FT readers.

Accelerating the development of life-saving treatments.

Developers can now integrate ChatGPT and Whisper models into their apps and products through our API.

GPT-3.5 Turbo, DALL·E and Whisper APIs are also generally available, and we are releasing a deprecation plan for older models of the Completions API, which will retire at the beginning of 2024.

Increasing enterprise support with more security features and controls, updates to our Assistants API, and tools to better manage costs.

Today's LLMs are susceptible to prompt injections, jailbreaks, and other attacks that allow adversaries to overwrite a model's original instructions with their own malicious prompts.

We are excited to announce our first office in Asia and we’re releasing a GPT-4 custom model optimized for the Japanese language.

Klarna is using AI to revolutionize personal shopping, customer service, and employee productivity.

We’re adding new features to help developers have more control over fine-tuning and announcing new ways to build custom models with OpenAI.

Harvey partners with OpenAI to build a custom-trained model for legal professionals.

We’re making it easier for people to experience the benefits of AI without needing to sign up

Oscar brings AI to health insurance, reducing costs and improving patient care.

We’re sharing lessons from a small scale preview of Voice Engine, a model for creating custom voices.

Zelma uses GPT-4 to make education data accessible.

OpenAI’s comment to the NTIA on open model weightsThis comment was submitted by OpenAI in response to NTIA’s March 2024 Request for Information on Dual-Use Foundation Models with Widely Available Weights.

Since we introduced Sora to the world last month, we’ve been working with artists to learn how Sora might aid in their creative process.

JetBrains uses OpenAI’s API to build its fastest-growing product ever.

Superhuman introduces a new era of email with OpenAI.

Salesforce integrates OpenAI’s enterprise-ready LLMs to transform customer applications.

Holiday Extras rolls out ChatGPT Enterprise across every team, boosting productivity by 500 hours weekly.

We have partnered with international news organizations Le Monde and Prisa Media to bring French and Spanish news content to ChatGPT.

Healthify collaborates with OpenAI to improve millions of lives with sustainable weight loss.

Dr. Sue Desmond-Hellmann, Nicole Seligman, Fidji Simo join; Sam Altman rejoins board

New board members named and enhancements to the governance structure introduced

Lifespan uses GPT-4 to radically improve health literacy and patient outcomes.

Paradigm uses OpenAI’s API to improve patient access to clinical trials.

Match Group uses ChatGPT Enterprise to spark creativity and impact.

We are dedicated to the OpenAI mission and have pursued it every step of the way.

We explore large-scale training of generative models on video data. Specifically, we train text-conditional diffusion models jointly on videos and images of variable durations, resolutions and aspect ratios. We leverage a transformer architecture that operates on spacetime patches of video and image latent codes. Our largest model, Sora, is capable of generating a minute of high fidelity video. Our results suggest that scaling video generation models is a promising path towards building general purpose simulators of the physical world.

We terminated accounts associated with state-affiliated threat actors. Our findings show our models offer only limited, incremental capabilities for malicious cybersecurity tasks.

We’re testing the ability for ChatGPT to remember things you discuss to make future chats more helpful. You’re in control of ChatGPT’s memory.

The National Institute of Standards and Technology (NIST) request for information related to its assignments under sections 4.1, 4.5, and 11 of the Executive Order Concerning Artificial Intelligence

We’re developing a blueprint for evaluating the risk that a large language model (LLM) could aid someone in creating a biological threat. In an evaluation involving both biology experts and students, we found that GPT-4 provides at most a mild uplift in biological threat creation accuracy. While this uplift is not large enough to be conclusive, our finding is a starting point for continued research and community deliberation.

We are launching a new generation of embedding models, new GPT-4 Turbo and moderation models, new API usage management tools, and soon, lower pricing on GPT-3.5 Turbo.

We funded 10 teams from around the world to design ideas and tools to collectively govern AI. We summarize the innovations, outline our learnings, and call for researchers and engineers to join us as we continue this work.

We’re working to prevent abuse, provide transparency on AI-generated content, and improve access to accurate voting information.

Digital Green uses OpenAI to increase farmer income.

We’re launching the GPT Store to help you find useful and popular custom versions of ChatGPT.

We’re launching a new ChatGPT plan for teams of all sizes, which provides a secure, collaborative workspace to get the most out of ChatGPT at work.

We support journalism, partner with news organizations, and believe The New York Times lawsuit is without merit.

WHOOP delivers personalized fitness and health coaching with GPT-4.

Summer Health reimagines pediatric doctor’s visits with OpenAI.

We’re launching $10M in grants to support technical research towards the alignment and safety of superhuman AI systems, including weak-to-strong generalization, interpretability, scalable oversight, and more.

We present a new research direction for superalignment, together with promising initial results: can we leverage the generalization properties of deep learning to control strong models with weak supervisors?

Axel Springer is the first publishing house globally to partner with us on a deeper integration of journalism in AI technologies.

Mira Murati as CTO, Greg Brockman returns as President. Read messages from CEO Sam Altman and board chair Bret Taylor.

Working together to create open-source and private datasets for AI training.

You can now create custom versions of ChatGPT that combine instructions, extra knowledge, and any combination of skills.

GPT-4 Turbo with 128K context and lower prices, the new Assistants API, GPT-4 Turbo with Vision, DALL·E 3 API, and more.

An Update for the UK AI Safety Summit

To support the safety of highly-capable AI systems, we are developing our approach to catastrophic risk preparedness, including building a Preparedness team and launching a challenge.

Together with Anthropic, Google, and Microsoft, we’re announcing the new Executive Director of the Frontier Model Forum and a new $10 million AI Safety Fund.

We developed a safety mitigation stack to ready DALL·E 3 for wider release and are sharing updates on our provenance research.

Ironclad uses GPT-4 to simplify the contract review process.

Typeform evolves online forms into dynamic and conversational data collection experiences with GPT-3.5 and GPT-4.

Retool uses GPT-4 to give businesses a fast, secure way to build AI-powered apps.

We are beginning to roll out new voice and image capabilities in ChatGPT. They offer a new, more intuitive type of interface by allowing you to have a voice conversation or show ChatGPT what you’re talking about.

We’re announcing an open call for the OpenAI Red Teaming Network and invite domain experts interested in improving the safety of OpenAI’s models to join our efforts.

We’re growing our presence in Europe with an office in Dublin, Ireland.

Developer registration for in-person attendance will open in the coming weeks and developers everywhere will be able to livestream the keynote.

We’re releasing a guide for teachers using ChatGPT in their classroom—including suggested prompts, an explanation of how ChatGPT works and its limitations, the efficacy of AI detectors, and bias.

Get enterprise-grade security & privacy and the most powerful version of ChatGPT yet.

OpenAI’s customers can leverage Scale’s AI expertise to customize our most advanced models.

Developers can now bring their own data to customize GPT-3.5 Turbo for their use cases.

The entire team has joined OpenAI.

We use GPT-4 for content policy development and content moderation decisions, enabling more consistent labeling, a faster feedback loop for policy refinement, and less involvement from human moderators.

We’re forming a new industry body to promote the safe and responsible development of frontier AI systems: advancing AI safety research, identifying best practices and standards, and facilitating information sharing among policymakers and industry.

OpenAI and other leading labs reinforce AI safety, security and trustworthiness through voluntary commitments.

We’re rolling out custom instructions to give you more control over how ChatGPT responds. Set your preferences, and ChatGPT will keep them in mind for all future conversations.

A new $5+ million partnership aims to explore ways the development of artificial intelligence (AI) can support a thriving, innovative local news field, and ensure local news organizations shape the future of this emerging technology.

Viable uses GPT-4 to analyze qualitative data at a revolutionary scale with unparalleled accuracy.

We are sharing what we learned from our conversations across 22 countries, and how we will be incorporating those insights moving forward.

We are excited to announce OpenAI’s first international expansion with a new office in London, United Kingdom.

The following is the written testimony of Sam Altman, Chief Executive Officer of OpenAI, before the U.S. Senate Committee on the Judiciary (Subcommittee on Privacy, Technology, & the Law).

The following are the Questions for the Record following Sam Altman's testimony before the U.S. Senate Committee on the Judiciary (Subcommittee on Privacy, Technology, & the Law).

We’re announcing updates including more steerable API models, function calling capabilities, longer context, and lower prices.

The National Telecommunications and Information Administration (NTIA) request for comments on AI Accountability policy.

Our goal is to facilitate the development of AI-powered cybersecurity capabilities for defenders through grants and other support.

We've trained a model to achieve a new state-of-the-art in mathematical problem solving by rewarding each correct step of reasoning (“process supervision”) instead of simply rewarding the correct final answer (“outcome supervision”). In addition to boosting performance relative to outcome supervision, process supervision also has an important alignment benefit: it directly trains the model to produce a chain-of-thought that is endorsed by humans.

Our nonprofit organization, OpenAI, Inc., is launching a program to award ten $100,000 grants to fund experiments in setting up a democratic process for deciding what rules AI systems should follow, within the bounds defined by the law.

Now is a good time to start thinking about the governance of superintelligence—future AI systems dramatically more capable than even AGI.

The ChatGPT app syncs your conversations, supports voice input, and brings our latest model improvements to your fingertips.

We use GPT-4 to automatically write explanations for the behavior of neurons in large language models and to score those explanations. We release a dataset of these (imperfect) explanations and scores for every neuron in GPT-2.

ChatGPT users can now turn off chat history, allowing you to choose which conversations can be used to train our models.

This initiative is essential to our commitment to develop safe and advanced AI. As we create technology and services that are secure, reliable, and trustworthy, we need your help.

Ensuring that AI systems are built, deployed, and used safely is critical to our mission.

An update on our findings, the actions we’ve taken, and technical details of the bug.

We’ve implemented initial support for plugins in ChatGPT. Plugins are tools designed specifically for language models with safety as a core principle, and help ChatGPT access up-to-date information, run computations, or use third-party services.

GPT-4 deepens the conversation on Duolingo.

How Iceland is using GPT-4 to preserve its language.

Stripe leverages GPT-4 to streamline user experience and combat fraud.

Be My Eyes uses GPT-4 to transform visual accessibility.

Khan Academy explores the potential for GPT-4 in a limited pilot program.

We’ve created GPT-4, the latest milestone in OpenAI’s effort in scaling up deep learning. GPT-4 is a large multimodal model (accepting image and text inputs, emitting text outputs) that, while less capable than humans in many real-world scenarios, exhibits human-level performance on various professional and academic benchmarks.

Our mission is to ensure that artificial general intelligence—AI systems that are generally smarter than humans—benefits all of humanity.

We’re clarifying how ChatGPT’s behavior is shaped and our plans for improving that behavior, allowing more user customization, and getting more public input into our decision-making in these areas.

We’re launching a pilot subscription plan for ChatGPT, a conversational AI that can chat with you, answer follow-up questions, and challenge incorrect assumptions.

We’re launching a classifier trained to distinguish between AI-written and human-written text.

We’re happy to announce that OpenAI and Microsoft are extending our partnership.

OpenAI researchers collaborated with Georgetown University’s Center for Security and Emerging Technology and the Stanford Internet Observatory to investigate how large language models might be misused for disinformation purposes. The collaboration included an October 2021 workshop bringing together 30 disinformation researchers, machine learning experts, and policy analysts, and culminated in a co-authored report building on more than a year of research. This report outlines the threats that language models pose to the information environment if used to augment disinformation campaigns and introduces a framework for analyzing potential mitigations. Read the full report here.

Using GPT-3 to deliver fast, nuanced insights from customer feedback.

Fine-tuning GPT-3 to power and scale done-for-you video creation.

Using GPT-3 to create the next generation of AI-powered characters.

Lilian Weng works on Applied AI Research at OpenAI.

We are excited to announce a new embedding model which is significantly more capable, cost effective, and simpler to use.

Christian Gibson is an engineer on the Supercomputing team at OpenAI.

We’ve trained a model called ChatGPT which interacts in a conversational way. The dialogue format makes it possible for ChatGPT to answer followup questions, admit its mistakes, challenge incorrect premises, and reject inappropriate requests.

Starting today, developers can begin building apps with the DALL·E API.

New users can start creating straight away. Lessons learned from deployment and improvements to our safety systems make wider availability possible.

We’ve trained and are open-sourcing a neural net called Whisper that approaches human level robustness and accuracy on English speech recognition.

Extend creativity and tell a bigger story with DALL·E images of any size.

We are improving our AI systems’ ability to learn from human feedback and to assist humans at evaluating AI. Our goal is to build a sufficiently aligned AI system that can help us solve all other alignment problems.

We are introducing a new and improved content moderation tool. The Moderation endpoint improves upon our previous content filter, and is available for free today to OpenAI API developers.

We’ll invite 1 million people from our waitlist over the coming weeks. Users can create with DALL·E using free credits that refill every month, and buy additional credits in 115-generation increments for $15.

Today, we are implementing a new technique so that DALL·E generates images of people that more accurately reflect the diversity of the world’s population.

As part of our DALL·E 2 research preview, more than 3,000 artists from more than 118 countries have incorporated DALL·E into their creative workflows. The artists in our early access group have helped us discover new uses for DALL·E and have served as key voices as we’ve made decisions about DALL·E’s features.

In order to share the magic of DALL·E 2 with a broad audience, we needed to reduce the risks associated with powerful image generation models. To this end, we put various guardrails in place to prevent generated images from violating our content policy.

We trained a neural network to play Minecraft by Video PreTraining (VPT) on a massive unlabeled video dataset of human Minecraft play, while using only a small amount of labeled contractor data. With fine-tuning, our model can learn to craft diamond tools, a task that usually takes proficient humans over 20 minutes (24,000 actions). Our model uses the native human interface of keypresses and mouse movements, making it quite general, and represents a step towards general computer-using agents.

We trained “critique-writing” models to describe flaws in summaries. Human evaluators find flaws in summaries much more often when shown our model’s critiques. Larger models are better at self-critiquing, with scale improving critique-writing more than summary-writing. This shows promise for using AI systems to assist human supervision of AI systems on difficult tasks.

Large neural networks are at the core of many recent advances in AI, but training them is a difficult engineering and research challenge which requires orchestrating a cluster of GPUs to perform a single synchronized calculation.

Cohere, OpenAI, and AI21 Labs have developed a preliminary set of best practices applicable to any organization developing or deploying large language models.

Codex is now powering 70 different applications across a variety of use cases through the OpenAI API.

Early users have created over 3 million images to date and helped us improve our safety processes. We’re excited to begin adding up to 1,000 new users from our waitlist each week.

We’re happy to announce several executive role changes that reflect our recent progress and will ensure continued momentum toward our next major milestones.

Goodhart’s law famously says: “When a measure becomes a target, it ceases to be a good measure.” Although originally from economics, it’s something we have to grapple with at OpenAI when figuring out how to optimize objectives that are difficult or costly to measure.

We’ve released new versions of GPT-3 and Codex which can edit or insert content into existing text, rather than just completing existing text.

Call for expressions of interest to study the economic impacts of large language models.

We describe our latest thinking in the hope of helping other AI developers address safety and misuse of deployed models.

We built a neural theorem prover for Lean that learned to solve a variety of challenging high-school olympiad problems, including problems from the AMC12 and AIME competitions, as well as two problems adapted from the IMO.

We’ve trained language models that are much better at following user intentions than GPT-3 while also making them more truthful and less toxic, using techniques developed through our alignment research. These InstructGPT models, which are trained with humans in the loop, are now deployed as the default language models on our API.

We are introducing embeddings, a new endpoint in the OpenAI API that makes it easy to perform natural language and code tasks like semantic search, clustering, topic modeling, and classification.

We’ve fine-tuned GPT-3 to more accurately answer open-ended questions using a text-based web browser.

Fine-tune with a single command.

As part of our effort to support and develop AI talent, we’re excited to announce the OpenAI Residency.

Wider availability made possible by safety progress.

We’ve trained a system that solves grade school math problems with nearly twice the accuracy of a fine-tuned GPT-3 model. It solves about 90% as many problems as real kids: a small sample of 9-12 year olds scored 60% on a test from our dataset, while our system scored 55% on those same problems.

Scaling human oversight of AI systems for tasks that are difficult to evaluate.

Today, we’re excited to announce the appointment of Helen Toner to our board of directors.

We’ve created an improved version of OpenAI Codex, our AI system that translates natural language to code, and we are releasing it through our API in private beta starting today.

We’re releasing Triton 1.0, an open-source Python-like programming language which enables researchers with no CUDA experience to write highly efficient GPU code—most of the time on par with what an expert would be able to produce.

Our latest research finds we can improve language model behavior with respect to specific behavioral values by fine-tuning on a small, curated dataset.

We’re proud to announce that the 2021 class of OpenAI Scholars has completed our six-month mentorship program and have produced an open-source research project with stipends and support from OpenAI.

OpenAI is committed to developing general-purpose artificial intelligence that benefits all humanity, and we believe that achieving our goal requires expertise in public policy as well as technology. So, we’re delighted to announce that Congressman Will Hurd has joined our board of directors.

Over 300 applications are delivering GPT-3–powered search, conversation, text completion, and other advanced AI features through our API.

We’ve discovered neurons in CLIP that respond to the same concept whether presented literally, symbolically, or conceptually. This may explain CLIP’s accuracy in classifying surprising visual renditions of concepts, and is also an important step toward understanding the associations and biases that CLIP and similar models learn.

We’ve scaled Kubernetes clusters to 7,500 nodes, producing a scalable infrastructure for large models like GPT-3, CLIP, and DALL·E, but also for rapid small-scale iterative research such as Scaling Laws for Neural Language Models.

We’re introducing a neural network called CLIP which efficiently learns visual concepts from natural language supervision. CLIP can be applied to any visual classification benchmark by simply providing the names of the visual categories to be recognized, similar to the “zero-shot” capabilities of GPT-2 and GPT-3.

We’ve trained a neural network called DALL·E that creates images from text captions for a wide range of concepts expressible in natural language.

It’s been a year of dramatic change and growth at OpenAI.

OpenAI has agreed to license GPT-3 to Microsoft for their own products and services.

We’ve applied reinforcement learning from human feedback to train language models that are better at summarization.

Our third class of OpenAI Scholars presented their final projects at virtual Demo Day, showcasing their research results from over the past five months.

We’re excited to announce that OpenAI is co-organizing two NeurIPS 2020 competitions with AIcrowd, Carnegie Mellon University, and DeepMind, using Procgen Benchmark and MineRL.

We find that, just as a large transformer model trained on language can generate coherent text, the same exact model trained on pixel sequences can generate coherent image completions and samples. By establishing a correlation between sample quality and image classification accuracy, we show that our best generative model also contains features competitive with top convolutional nets in the unsupervised setting.

We’re releasing an API for accessing new AI models developed by OpenAI.

We’re releasing an analysis showing that since 2012 the amount of compute needed to train a neural net to the same performance on ImageNet classification has been decreasing by a factor of 2 every 16 months. Compared to 2012, it now takes 44 times less compute to train a neural network to the level of AlexNet (by contrast, Moore’s Law would yield an 11x cost improvement over this period). Our results suggest that for AI tasks with high levels of recent investment, algorithmic progress has yielded more gains than classical hardware efficiency.

We’re introducing Jukebox, a neural net that generates music, including rudimentary singing, as raw audio in a variety of genres and artist styles. We’re releasing the model weights and code, along with a tool to explore the generated samples.

We’ve contributed to a multi-stakeholder report by 58 co-authors at 30 organizations, including the Centre for the Future of Intelligence, Mila, Schwartz Reisman Institute for Technology and Society, Center for Advanced Study in the Behavioral Sciences, and Center for Security and Emerging Technologies. This report describes 10 mechanisms to improve the verifiability of claims made about AI systems. Developers can use these tools to provide evidence that AI systems are safe, secure, fair, or privacy-preserving. Users, policymakers, and civil society can use these tools to evaluate AI development processes.

We’re introducing OpenAI Microscope, a collection of visualizations of every significant layer and neuron of eight vision “model organisms” which are often studied in interpretability. Microscope makes it easier to analyze the features that form inside these neural networks, and we hope it will help the research community as we move towards understanding these complicated systems.

We are standardizing OpenAI’s deep learning framework on PyTorch.

We show that the double descent phenomenon occurs in CNNs, ResNets, and transformers: performance first improves, then gets worse, and then improves again with increasing model size, data size, or training time. This effect is often avoided through careful regularization. While this behavior appears to be fairly universal, we don’t yet fully understand why it happens, and view further study of this phenomenon as an important research direction.

We’re releasing Procgen Benchmark, 16 simple-to-use procedurally-generated environments which provide a direct measure of how quickly a reinforcement learning agent learns generalizable skills.

We’re releasing Safety Gym, a suite of environments and tools for measuring progress towards reinforcement learning agents that respect safety constraints while training.

As the final model release of GPT-2’s staged release, we’re releasing the largest version (1.5B parameters) of GPT-2 along with code and model weights to facilitate detection of outputs of GPT-2 models. While there have been larger language models released since August, we’ve continued with our original staged release plan in order to provide the community with a test case of a full staged release process. We hope that this test case will be useful to developers of future powerful models, and we’re actively continuing the conversation with the AI community on responsible publication.

We’ve trained a pair of neural networks to solve the Rubik’s Cube with a human-like robot hand. The neural networks are trained entirely in simulation, using the same reinforcement learning code as OpenAI Five paired with a new technique called Automatic Domain Randomization (ADR). The system can handle situations it never saw during training, such as being prodded by a stuffed giraffe. This shows that reinforcement learning isn’t just a tool for virtual tasks, but can solve physical-world problems requiring unprecedented dexterity.

We are now accepting applications for our third class of OpenAI Scholars.

We’ve fine-tuned the 774M parameter GPT-2 language model using human feedback for various tasks, successfully matching the preferences of the external human labelers, though those preferences did not always match our own. Specifically, for summarization tasks the labelers preferred sentences copied wholesale from the input (we’d only asked them to ensure accuracy), so our models learned to copy. Summarization required 60k human labels; simpler tasks which continue text in various styles required only 5k. Our motivation is to move safety techniques closer to the general task of “machines talking to humans,” which we believe is key to extracting information about human values.

We’ve observed agents discovering progressively more complex tool use while playing a simple game of hide-and-seek. Through training in our new simulated hide-and-seek environment, agents build a series of six distinct strategies and counterstrategies, some of which we did not know our environment supported. The self-supervised emergent complexity in this simple environment further suggests that multi-agent co-adaptation may one day produce extremely complex and intelligent behavior.

We’ve developed a method to assess whether a neural network classifier can reliably defend against adversarial attacks not seen during training. Our method yields a new metric, UAR (Unforeseen Attack Robustness), which evaluates the robustness of a single model against an unanticipated attack, and highlights the need to measure performance across a more diverse range of unforeseen attacks.

We’re releasing the 774 million parameter GPT-2 language model after the release of our small 124M model in February, staged release of our medium 355M model in May, and subsequent research with partners and the AI community into the model’s potential for misuse and societal benefit. We’re also releasing an open-source legal agreement to make it easier for organizations to initiate model-sharing partnerships with each other, and are publishing a technical report about our experience in coordinating with the wider AI research community on publication norms.

At OpenAI, each Thursday is Learning Day: a day where employees have the option to self-study technical skills that will make them better at their job but which aren’t being learned from daily work.

Microsoft is investing $1 billion in OpenAI to support us building artificial general intelligence (AGI) with widely distributed economic benefits. We’re partnering to develop a hardware and software platform within Microsoft Azure which will scale to AGI. We’ll jointly develop new Azure AI supercomputing technologies, and Microsoft will become our exclusive cloud provider—so we’ll be working hard together to further extend Microsoft Azure’s capabilities in large-scale AI systems.

We’ve written a policy research paper identifying four strategies that can be used today to improve the likelihood of long-term industry cooperation on safety norms in AI: communicating risks and benefits, technical collaboration, increased transparency, and incentivizing standards. Our analysis shows that industry cooperation on safety will be instrumental in ensuring that AI systems are safe and beneficial, but competitive pressures could lead to a collective action problem, potentially causing AI companies to under-invest in safety. We hope these strategies will encourage greater cooperation on the safe development of AI and lead to better global outcomes of AI.

We hosted the first OpenAI Robotics Symposium on April 27, 2019.

Our second class of OpenAI Scholars has concluded, with all eight scholars producing an exciting final project showcased at Scholars Demo Day at OpenAI.

Our second class of OpenAI Fellows has wrapped up, with each Fellow going from a machine learning beginner to core OpenAI contributor in the course of a 6-month apprenticeship. We are currently reviewing applications on a rolling basis for our next round of OpenAI Fellows Summer 2019.

We’ve created MuseNet, a deep neural network that can generate 4-minute musical compositions with 10 different instruments, and can combine styles from country to Mozart to the Beatles. MuseNet was not explicitly programmed with our understanding of music, but instead discovered patterns of harmony, rhythm, and style by learning to predict the next token in hundreds of thousands of MIDI files. MuseNet uses the same general-purpose unsupervised technology as GPT-2, a large-scale transformer model trained to predict the next token in a sequence, whether audio or text.

We’ve developed the Sparse Transformer, a deep neural network which sets new records at predicting what comes next in a sequence—whether text, images, or sound. It uses an algorithmic improvement of the attention mechanism to extract patterns from sequences 30x longer than possible previously.

OpenAI Five is the first AI to beat the world champions in an esports game, having won two back-to-back games versus the world champion Dota 2 team, OG, at Finals this weekend. Both OpenAI Five and DeepMind’s AlphaStar had previously beaten good pros privately but lost their live pro matches, making this also the first time an AI has beaten esports pros on livestream.

We’ll be holding our final live event for OpenAI Five at 11:30am PT on April 13.

We’ve made progress towards stable and scalable training of energy-based models (EBMs) resulting in better sample quality and generalization ability than existing models. Generation in EBMs spends more compute to continually refine its answers and doing so can generate samples competitive with GANs at low temperatures, while also having mode coverage guarantees of likelihood-based models. We hope these findings stimulate further research into this promising class of models.

Our class of eight scholars (out of 550 applicants) brings together collective expertise in literature, philosophy, cell biology, statistics, economics, quantum physics, and business innovation.

We’ve created OpenAI LP, a new “capped-profit” company that allows us to rapidly increase our investments in compute and talent while including checks and balances to actualize our mission.

We’ve created activation atlases (in collaboration with Google researchers), a new technique for visualizing what interactions between neurons can represent. As AI systems are deployed in increasingly sensitive contexts, having a better understanding of their internal decision-making processes will let us identify weaknesses and investigate failures.

We’re releasing a Neural MMO, a massively multiagent game environment for reinforcement learning agents. Our platform supports a large, variable number of agents within a persistent and open-ended task. The inclusion of many agents and species leads to better exploration, divergent niche formation, and greater overall competence.

On February 2, we held our first Spinning Up Workshop as part of our new education initiative at OpenAI.

We’ve written a paper arguing that long-term AI safety research needs social scientists to ensure AI alignment algorithms succeed when actual humans are involved. Properly aligning advanced AI systems with human values requires resolving many uncertainties related to the psychology of human rationality, emotion, and biases. The aim of this paper is to spark further collaboration between machine learning and social science researchers, and we plan to hire social scientists to work on this full time at OpenAI.

We’ve trained a large-scale unsupervised language model which generates coherent paragraphs of text, achieves state-of-the-art performance on many language modeling benchmarks, and performs rudimentary reading comprehension, machine translation, question answering, and summarization—all without task-specific training.

Our first cohort of OpenAI Fellows has concluded, with each Fellow going from a machine learning beginner to core OpenAI contributor in the course of a 6-month apprenticeship.

We’ve discovered that the gradient noise scale, a simple statistical metric, predicts the parallelizability of neural network training on a wide range of tasks. Since complex tasks tend to have noisier gradients, increasingly large batch sizes are likely to become useful in the future, removing one potential limit to further growth of AI systems. More broadly, these results show that neural network training need not be considered a mysterious art, but can be rigorized and systematized.

We’re releasing CoinRun, a training environment which provides a metric for an agent’s ability to transfer its experience to novel situations and has already helped clarify a longstanding puzzle in reinforcement learning. CoinRun strikes a desirable balance in complexity: the environment is simpler than traditional platformer games like Sonic the Hedgehog but still poses a worthy generalization challenge for state of the art algorithms.

We’re releasing Spinning Up in Deep RL, an educational resource designed to let anyone learn to become a skilled practitioner in deep reinforcement learning. Spinning Up consists of crystal-clear examples of RL code, educational exercises, documentation, and tutorials.

We’ve developed an energy-based model that can quickly learn to identify and generate instances of concepts, such as near, above, between, closest, and furthest, expressed as sets of 2d points. Our model learns these concepts after only five demonstrations. We also show cross-domain transfer: we use concepts learned in a 2d particle environment to solve tasks on a 3-dimensional physics-based robot.

We’ve developed Random Network Distillation (RND), a prediction-based method for encouraging reinforcement learning agents to explore their environments through curiosity, which for the first time exceeds average human performance on Montezuma’s Revenge.

We’re proposing an AI safety technique called iterated amplification that lets us specify complicated behaviors and goals that are beyond human scale, by demonstrating how to decompose a task into simpler sub-tasks, rather than by providing labeled data or a reward function. Although this idea is in its very early stages and we have only completed experiments on simple toy algorithmic domains, we’ve decided to present it in its preliminary state because we think it could prove to be a scalable approach to AI safety.

We are now accepting applications for our second cohort of OpenAI Scholars, a program where we provide 6–10 stipends and mentorship to individuals from underrepresented groups to study deep learning full-time for 3 months and open-source a project.

We are now accepting applications for OpenAI Fellows and Interns for 2019.

Our first cohort of OpenAI Scholars has now completed the program.

OpenAI Five lost two games against top Dota 2 players at The International in Vancouver this week, maintaining a good chance of winning for the first 20–35 minutes of both games.

Yesterday, OpenAI Five won a best-of-three against a team of 99.95th percentile Dota players: Blitz, Cap, Fogged, Merlini, and MoonMeander—four of whom have played Dota professionally—in front of a live audience and 100,000 concurrent livestream viewers.

We’ve trained a human-like robot hand to manipulate physical objects with unprecedented dexterity.

Our first class of OpenAI Scholars is underway, and you can now follow along as this group of experienced software developers becomes machine learning practitioners.

The OpenAI Five Benchmark match is now over!

We introduce Glow, a reversible generative model which uses invertible 1x1 convolutions. It extends previous work on reversible generative models and simplifies the architecture. Our model can generate realistic high resolution images, supports efficient sampling, and discovers features that can be used to manipulate attributes of data. We’re releasing code for the model and an online visualization tool so people can explore and build on these results.

We’ve trained an agent to achieve a high score of 74,500 on Montezuma’s Revenge from a single human demonstration, better than any previously published result. Our algorithm is simple: the agent plays a sequence of games starting from carefully chosen states from the demonstration, and learns from them by optimizing the game score using PPO, the same reinforcement learning algorithm that underpins OpenAI Five.

Our team of five neural networks, OpenAI Five, has started to defeat amateur human teams at Dota 2.

The first run of our Retro Contest—exploring the development of algorithms that can generalize from previous experience—is now complete.

We’ve obtained state-of-the-art results on a suite of diverse language tasks with a scalable, task-agnostic system, which we’re also releasing. Our approach is a combination of two existing ideas: transformers and unsupervised pre-training. These results provide a convincing example that pairing supervised learning methods with unsupervised pre-training works very well; this is an idea that many have explored in the past, and we hope our result motivates further research into applying this idea on larger and more diverse datasets.

We’re now accepting applications for the next cohort of OpenAI Fellows, a program which offers a compensated 6-month apprenticeship in AI research at OpenAI.

We’re releasing the full version of Gym Retro, a platform for reinforcement learning research on games. This brings our publicly-released game count from around 70 Atari games and 30 Sega games to over 1,000 games across a variety of backing emulators. We’re also releasing the tool we use to add new games to the platform.

We’re releasing an analysis showing that since 2012, the amount of compute used in the largest AI training runs has been increasing exponentially with a 3.4-month doubling time (by comparison, Moore’s Law had a 2-year doubling period)[^footnote-correction]. Since 2012, this metric has grown by more than 300,000x (a 2-year doubling period would yield only a 7x increase). Improvements in compute have been a key component of AI progress, so as long as this trend continues, it’s worth preparing for the implications of systems far outside today’s capabilities.

We’re proposing an AI safety technique which trains agents to debate topics with one another, using a human to judge who wins.

We’re releasing an experimental metalearning approach called Evolved Policy Gradients, a method that evolves the loss function of learning agents, which can enable fast training on novel tasks. Agents trained with EPG can succeed at basic tasks at test time that were outside their training regime, like learning to navigate to an object on a different side of the room from where it was placed during training.

We’re launching a transfer learning contest that measures a reinforcement learning algorithm’s ability to generalize from previous experience.

On March 3rd, we hosted our first hackathon with 100 members of the artificial intelligence community.

We’ve developed a simple meta-learning algorithm called Reptile which works by repeatedly sampling a task, performing stochastic gradient descent on it, and updating the initial parameters towards the final parameters learned on that task. Reptile is the application of the Shortest Descent algorithm to the meta-learning setting, and is mathematically similar to first-order MAML (which is a version of the well-known MAML algorithm) that only needs black-box access to an optimizer such as SGD or Adam, with similar computational efficiency and performance.

We’re providing 6–10 stipends and mentorship to individuals from underrepresented groups to study deep learning full-time for 3 months and open-source a project.

We’re releasing eight simulated robotics environments and a Baselines implementation of Hindsight Experience Replay, all developed for our research over the past year. We’ve used these environments to train models which work on physical robots. We’re also releasing a set of requests for robotics research.

Come to OpenAI’s office in San Francisco’s Mission District for talks and a hackathon on Saturday, March 3rd.

We’ve co-authored a paper that forecasts how malicious actors could misuse AI technology, and potential ways we can prevent and mitigate these threats. This paper is the outcome of almost a year of sustained work with our colleagues at the Future of Humanity Institute, the Centre for the Study of Existential Risk, the Center for a New American Security, the Electronic Frontier Foundation, and others.

We’re excited to welcome new donors to OpenAI.

We’ve designed a method that encourages AIs to teach each other with examples that also make sense to humans. Our approach automatically selects the most informative examples to teach a concept—for instance, the best images to describe the concept of dogs—and experimentally we found our approach to be effective at teaching both AIs

We’ve built a system for automatically figuring out which object is meant by a word by having a neural network decide if the word belongs to each of about 100 automatically-discovered “types” (non-exclusive categories).

We’re releasing a new batch of seven unsolved problems which have come up in the course of our research at OpenAI.

We’re releasing highly-optimized GPU kernels for an underexplored class of neural network architectures: networks with block-sparse weights. Depending on the chosen sparsity, these kernels can run orders of magnitude faster than cuBLAS or cuSPARSE. We’ve used them to attain state-of-the-art results in text sentiment analysis and generative modeling of text and images.

We’ve developed a hierarchical reinforcement learning algorithm that learns high-level actions useful for solving a range of tasks, allowing fast solving of tasks requiring thousands of timesteps. Our algorithm, when applied to a set of navigation problems, discovers a set of high-level actions for walking and crawling in different directions, which enables the agent to master new navigation tasks quickly.

Our latest robotics techniques allow robot controllers, trained entirely in simulation and deployed on physical robots, to react to unplanned changes in the environment as they solve simple tasks. That is, we’ve used these techniques to build closed-loop systems rather than open-loop ones as before.

We show that for the task of simulated robot wrestling, a meta-learning agent can learn to quickly defeat a stronger non-meta-learning agent, and also show that the meta-learning agent can adapt to physical malfunction.

We’ve found that self-play allows simulated AIs to discover physical skills like tackling, ducking, faking, kicking, catching, and diving for the ball, without explicitly designing an environment with these skills in mind. Self-play ensures that the environment is always the right difficulty for an AI to improve. Taken alongside our Dota 2 self-play results, we have increasing confidence that self-play will be a core part of powerful AI systems in the future.

We’re releasing an algorithm which accounts for the fact that other agents are learning too, and discovers self-interested yet collaborative strategies like tit-for-tat in the iterated prisoner’s dilemma. This algorithm, Learning with Opponent-Learning Awareness (LOLA), is a small step towards agents that model other minds.

We’re releasing two new OpenAI Baselines implementations: ACKTR and A2C. A2C is a synchronous, deterministic variant of Asynchronous Advantage Actor Critic (A3C) which we’ve found gives equal performance. ACKTR is a more sample-efficient reinforcement learning algorithm than TRPO and A2C, and requires only slightly more computation than A2C per update.

Our Dota 2 result shows that self-play can catapult the performance of machine learning systems from far below human level to superhuman, given sufficient compute. In the span of a month, our system went from barely matching a high-ranked player to beating the top pros and has continued to improve since then. Supervised deep learning systems can only be as good as their training datasets, but in self-play systems, the available data improves automatically as the agent gets better.

We’ve created a bot which beats the world’s top professionals at 1v1 matches of Dota 2 under standard tournament rules. The bot learned the game from scratch by self-play, and does not use imitation learning or tree search. This is a step towards building AI systems which accomplish well-defined goals in messy, complicated situations involving real humans.

RL-Teacher is an open-source implementation of our interface to train AIs via occasional human feedback rather than hand-crafted reward functions. The underlying technique was developed as a step towards safe AI systems, but also applies to reinforcement learning problems with rewards that are hard to specify.

We’ve found that adding adaptive noise to the parameters of reinforcement learning algorithms frequently boosts performance. This exploration method is simple to implement and very rarely decreases performance, so it’s worth trying on any problem.

We’re releasing a new class of reinforcement learning algorithms, Proximal Policy Optimization (PPO), which perform comparably or better than state-of-the-art approaches while being much simpler to implement and tune. PPO has become the default reinforcement learning algorithm at OpenAI because of its ease of use and good performance.

We’ve created images that reliably fool neural network classifiers when viewed from varied scales and perspectives. This challenges a claim from last week that self-driving cars would be hard to trick maliciously since they capture images from multiple scales, angles, perspectives, and the like.

We’re open-sourcing a high-performance Python library for robotic simulation using the MuJoCo engine, developed over our past year of robotics research.

One step towards building safe AI systems is to remove the need for humans to write goal functions, since using a simple proxy for a complex goal, or getting the complex goal a bit wrong, can lead to undesirable and even dangerous behavior. In collaboration with DeepMind’s safety team, we’ve developed an algorithm which can infer what humans want by being told which of two proposed behaviors is better.

Multiagent environments where agents compete for resources are stepping stones on the path to AGI. Multiagent environments have two useful properties: first, there is a natural curriculum—the difficulty of the environment is determined by the skill of your competitors (and if you’re competing against clones of yourself, the environment exactly matches your skill level). Second, a multiagent environment has no stable equilibrium: no matter how smart an agent is, there’s always pressure to get smarter. These environments have a very different feel from traditional environments, and it’ll take a lot more research before we become good at them.

We’re open-sourcing OpenAI Baselines, our internal effort to reproduce reinforcement learning algorithms with performance on par with published results. We’ll release the algorithms over upcoming months; today’s release includes DQN and three of its variants.

We’ve created a robotics system, trained entirely in simulation and deployed on a physical robot, which can learn a new task after seeing it done once.

We are releasing Roboschool: open-source software for robot simulation, integrated with OpenAI Gym.

We’ve developed an unsupervised system which learns an excellent representation of sentiment, despite being trained only to predict the next character in the text of Amazon reviews.

We’ve created the world’s first Spam-detecting AI trained entirely in simulation and deployed on a physical robot.

We’ve discovered that evolution strategies (ES), an optimization technique that’s been known for decades, rivals the performance of standard reinforcement learning (RL) techniques on modern RL benchmarks (e.g. Atari/MuJoCo), while overcoming many of RL’s inconveniences.

We’re excited to support today’s launch of Distill, a new kind of journal aimed at excellent communication of machine learning results (novel or existing).

In this post we’ll outline new OpenAI research in which agents develop their own language.

Adversarial examples are inputs to machine learning models that an attacker has intentionally designed to cause the model to make a mistake; they’re like optical illusions for machines. In this post we’ll show how adversarial examples work across different mediums, and will discuss why securing systems against them can be difficult.

The OpenAI team is now 45 people. Together, we’re pushing the frontier of AI capabilities—whether by validating novel ideas, creating new software systems, or deploying machine learning on robots.

Reinforcement learning algorithms can break in surprising, counterintuitive ways. In this post we’ll explore one failure mode, which is where you misspecify your reward function.

We’re releasing Universe, a software platform for measuring and training an AI’s general intelligence across the world’s supply of games, websites and other applications.

We’re working with Microsoft to start running most of our large-scale experiments on Azure.

Last week we hosted over a hundred and fifty AI practitioners in our offices for our first self-organizing conference on machine learning.

Deep learning is an empirical science, and the quality of a group’s infrastructure is a multiplier on progress. Fortunately, today’s open-source ecosystem makes it possible for anyone to build great deep learning infrastructure.

The latest information about the Unconference is now available at the Unconference wiki, which will be periodically updated with more information for attendees.

We’ve hired more great people to help us achieve our goals. Welcome, everyone!

Impactful scientific work requires working on the right problems—problems which are not just interesting, but whose solutions matter.

We (along with researchers from Berkeley and Stanford) are co-authors on today’s paper led by Google Brain researchers, Concrete Problems in AI Safety. The paper explores many research problems around ensuring that modern machine learning systems operate as intended.

OpenAI’s mission is to build safe AI, and ensure AI’s benefits are as widely and evenly distributed as possible.

This post describes four projects that share a common theme of enhancing or using generative models, a branch of unsupervised learning techniques in machine learning. In addition to describing our work, this post will tell you a bit more about generative models: what they are, why they are important, and where they might be going.

We’d like to welcome the latest set of team members to OpenAI (and we’re still hiring!)

We’re releasing the public beta of OpenAI Gym, a toolkit for developing and comparing reinforcement learning (RL) algorithms. It consists of a growing suite of environments (from simulated robots to Atari games), and a site for comparing and reproducing results.

We have two more team updates.

We've had some fantastic people join over the past few months (and we're still hiring). Welcome, everyone!

OpenAI is a non-profit artificial intelligence research company. Our goal is to advance digital intelligence in the way that is most likely to benefit humanity as a whole, unconstrained by a need to generate financial return. Since our research is free from financial obligations, we can better focus on a positive human impact.