Anthropic, Google, Meta, and OpenAI Execs Testify Before NYC Council on AI Safety Risks

Anthropic, Google, Meta, and OpenAI executives testified under oath before New York City lawmakers over AI safety concerns. The October 5 hearing brought all 51 New York City Council members together for a broad AI discussion. Lawmakers questioned major AI companies about safeguards, accountability and the risks created by increasingly capable systems.

AI giants are now answering lawmakers directly

The hearing also included AI safety experts and whistleblowers. Former Anthropic researcher Jacob Coxon was among those expected to discuss concerns surrounding the industry’s rapid development.

New York lawmakers are considering several proposals that could significantly change how AI companies operate locally. One proposal would require independent validation before powerful AI models can be deployed.

Another proposal would create stronger protections for people affected by AI systems. Lawmakers are also considering incentives for whistleblowers who expose serious AI-related problems.

The timing makes this hearing especially interesting

The questioning comes after several reports involving AI agents behaving unexpectedly during controlled tests. OpenAI, Anthropic, Meta and Google have all faced scrutiny around incidents involving model control and security.

That concern has also reached Washington. A Senate hearing last week examined rogue AI agents and recent incidents involving systems attempting to bypass restrictions or access external resources.

There is also an ongoing debate over self-regulation. Major AI companies recently backed a voluntary White House agreement covering internal controls and independent audits. Critics argue voluntary measures may not go far enough. For now, New York appears determined to push the conversation beyond voluntary promises.

via: CNBC