Entity
OpenAI-Anthropic-Google Secret AI Safety Collaboration
OpenAI's global head of policy Chris Lehane disclosed that OpenAI, Anthropic, and Google DeepMind have been secretly collaborating on a shared AI safety framework, an unusual instance of cooperation among fiercely competitive frontier AI labs.
This page was written and is updated by the paper from the source articles listed below. Importance is how central the topic is to current coverage; confidence is how well-sourced the page is; mentions counts the articles that fed it. How the wiki works →
## Overview
OpenAI Group PBC's global head of policy Chris Lehane said the company has been working closely with rivals Anthropic PBC and Google LLC's DeepMind research lab to try to devise a framework ensuring AI safety, revealing previously undisclosed cross-lab cooperation (SiliconANGLE, September 16).
## Key Facts
- Lehane made the disclosure at a press conference, stating: "It's better to try to work together to prioritize safety" (SiliconANGLE, September 16).
- The collaboration reportedly involves OpenAI, Anthropic, and Google DeepMind jointly working on an AI safety framework (SiliconANGLE, September 16).
- The cooperation was described as having been conducted secretly prior to this disclosure (SiliconANGLE, September 16).
## Strategic Context
This is a notable development given the intense competitive rivalry between these three frontier AI labs across model releases, enterprise deals, and talent poaching. Cross-lab safety coordination, if substantive, could shape future self-regulatory frameworks and potentially preempt or inform government AI safety legislation, connecting to broader threads including OpenAI's federal AI governance framework proposal, Anthropic's AI export controls advocacy, and the Steve Bannon/Bernie Sanders bipartisan AI safety coalition.
## Why It Matters
For policymakers and industry watchers, secret coordination among the three dominant frontier AI labs raises both promise (coordinated safety standards) and concern (potential antitrust or regulatory capture implications). This is likely to be referenced in future debates over AI self-regulation versus binding government oversight, and in scrutiny of whether such coordination constitutes anticompetitive behavior among dominant AI labs.