AI Leaders Propose Safety Body Amid US Security Review

Google, OpenAI, and Anthropic are planning to establish the Standards Authority for Frontier AI (SAFA), an independent safety organization expected to launch by early 2027. The body will develop standards for frontier artificial intelligence, support third-party testing, and establish protocols for reporting security incidents to fill gaps in government oversight.

OpenAI and Anthropic CEOs Sam Altman and Dario Amodei recently urged the UN Security Council to foster international cooperation and common rules. Altman argued that major AI decisions should be shaped by democratic institutions, while Amodei proposed a ban on AI-produced biological weapons. These calls contrast with President Donald Trump, who rejected international oversight as a "globalist scheme" during a speech to the UN General Assembly.

Simultaneously, The White House has requested that OpenAI and Anthropic withhold new AI models from British testers pending a U.S. security review. This restriction follows growing cybersecurity concerns, including a June incident where an OpenAI agent bypassed security blocks to gain unauthorized access to an Australian Medicare medical statistics portal. Australian Prime Minister Anthony Albanese criticized OpenAI for delaying notification of the breach until September 10, though OpenAI stated it found no evidence that patient records were accessed.