Trump Unveils Voluntary AI Safety Pact With Six Tech Giants, No Penalties for Violations

On September 29, 2026, President Trump and House Speaker Mike Johnson announced a voluntary AI safety accord at the White House, joined by Meta, Google, OpenAI, Nvidia, xAI and Anthropic. The companies agreed to internal risk reviews, external audits and board-level oversight committees, but the pact imposes no penalties for noncompliance, no mandatory public disclosure of audit findings, and leaves each company discretion over how to implement safeguards. Trump said the firms "are really going to be policing each other," while Johnson called it "a statement of principles."

The arrangement is notable because several signatories had previously called for binding federal rules rather than self-regulation. OpenAI's Chris Lehane had urged mandatory national AI safety standards weeks earlier, and executives at Anthropic and Google DeepMind had separately pushed for government oversight bodies. Choosing a voluntary framework over legislation lets the administration avoid formal rulemaking and congressional involvement while presenting itself as addressing safety concerns researchers say could carry severe risk.

The accord carries no enforcement mechanism. The White House has said safeguards "might eventually become mandatory" through future legislation, but none is pending, and alternative binding proposals in Congress have not advanced.