Trump and tech leaders sign "morally binding" AI safety accord for industry self-regulation

The accord outlines four rules for participating companies to ensure frontier AI models behave as intended:

1. Implement robust internal controls to monitor capabilities and alignment regarding cybersecurity, biosecurity, and chemical threats.
2. Empower an internal team to ensure all monitoring, detection, and remediation processes are functioning.
3. Partner with an independent external auditor or evaluator to carry out independent assessments.
4. Designate an independent committee of the board of directors to oversee safety teams and audit reports.

The agreement states these measures aim to provide confidence to users and the public, noting that these steps may eventually be codified into laws or regulations. However, the rules currently lack legal weight and do not mention penalties for violations, relying instead on voluntary compliance. This comes following recent reports of AI models from Google, OpenAI, and Anthropic involving unauthorized access to other companies and government websites.

During the announcement, President Trump stated that the industry is working to ensure technology is developed responsibly and safely, suggesting that "the smartest people in the world" are now watching over each other instead of immediate federal regulation. He also referred to the technology as "Super Intelligence" (SI) moving forward.