When Bots Agree: OpenAI and Anthropic Hold Hands in Washington 🤝

OpenAI and Anthropic are jointly lobbying Washington for a federal review framework that would evaluate advanced AI models before public release, according to a report. The proposal, submitted to the White House Office of Science and Technology Policy, calls for consistent AI safety standards applied across all major developers rather than rules negotiated company by company.

The two firms, which compete directly in the commercial AI market, are coordinating on policy as US lawmakers prepare to define new evaluation benchmarks for frontier systems. Their joint effort covers pre-deployment testing, third-party audits, and disclosure requirements for models trained above specified compute thresholds.

National security concerns, particularly around advanced AI development in China, are cited as a driver of the alliance. Both companies have warned that fragmented regulation could leave the United States vulnerable to adversaries while slowing domestic innovation, and they are urging the federal government to adopt a unified framework before state-level rules diverge.

The two firms have previously clashed publicly over AI safety philosophy, with Anthropic positioning itself as a more cautious actor and OpenAI advocating faster deployment under guardrails. Their convergence in Washington marks a notable shift in tone, even as their product roadmaps continue to compete.

The proposal lands as the White House considers an executive order on AI and as congressional committees draft legislation on frontier model oversight. The Department of Commerce, through the National Telecommunications and Information Administration, has also solicited input on accountability measures for the most capable systems.

OpenAI and Anthropic said the framework would apply to any developer meeting the defined compute or capability thresholds, regardless of whether the developer is based in the United States or abroad. Neither company disclosed a specific timeline for federal action, and the White House has not publicly committed to a draft timeline for the proposed standards.