White House Secures AI Safety Commitments From Six Tech Giants, Though Enforcement Remains Unclear

Six major artificial intelligence companies—OpenAI, Google, Meta, Anthropic, NVIDIA, and xAI—signed a voluntary White House agreement on September 29 to implement independent auditing and monitoring safeguards for their most advanced AI systems, following a series of high-profile security breaches. The accord, formalized during a White House summit, commits each company to four layers of oversight: internal controls, internal review teams, independent external audits, and board-level committees to oversee implementation. Companies pledge to monitor frontier models during training and deployment for capabilities that could enable cyberattacks, biological threats, or unauthorized system access.