OpenAI opens internal systems to four outside safety groups, including Goodfire and METR
OpenAI has invited four outside organisations into its internal systems for what president Greg Brockman called its "first round of embedded assessments", 36kr reports: METR, Apollo Research, Transluce and Goodfire. Each gets scoped access to internal systems and unreleased models.
Interpretability startup Goodfire said in a blog post that its researchers will embed within OpenAI and use its interpretability platform to monitor training runs, debug undesired behaviour and improve alignment, work it calls "model forensics". OpenAI researchers are the main audience for its findings, but Goodfire says it will also publish learnings as the collaboration progresses.
According to 36kr, METR will first review the largest frontier reinforcement learning training run OpenAI has launched. Apollo Research is sending a six-person team on site for four months, and Transluce will begin a six-week review in early December with access to unreleased models and raw chains of thought.
We hope you enjoyed this article
Consider subscribing to one of our newsletters like AI Policy Brief or Daily AI Brief.
Also, consider following us on social media:
More from AI Safety
Nov 16 EU science and ethics advisers say AI needs rigorous safeguards and rules that keep pace Nov 16 18 countries and AI firms including Anthropic and OpenAI sign statement on kill switches and whistleblower protection Nov 16 Transluce embeds with OpenAI for a six-week review of its alignment evaluations Nov 16 California sues OpenAI over its AI agents' cyberattack on Hugging Face Nov 16 OpenAI embeds six Apollo Research evaluators to test whether models try to subvert safety controlsAI Policy Brief
Weekly report on AI regulations, safety standards, government policies, and compliance requirements worldwide.
Market report
2025 Generative AI in Professional Services Report
Thomson Reuters
This report by Thomson Reuters explores the integration and impact of generative AI technologies, such as ChatGPT and Microsoft Copilot, within the professional services sector. It highlights the growing adoption of GenAI tools across industries like legal, tax, accounting, and government, and discusses the challenges and opportunities these technologies present. The report also examines professionals' perceptions of GenAI and the need for strategic integration to maximize its value.
Read moreYou may also like
OpenAI considered buying Anthropic-backed Goodfire for more than $5B, report says
Transluce embeds with OpenAI for a six-week review of its alignment evaluations
OpenAI Expands Monitoring for Frontier Model Misalignment
External test found gaps in OpenAI agent monitoring
Goodfire trains AI models to avoid misleading shortcuts
Daily AI Brief: the AI news that matters, in your inbox.