White House and Tech Giants Sign Voluntary 'Superintelligence' Safety Pact

Overview

• The White House hosted about 20 tech executives and published the 'White House Agreement on Superintelligence' that companies signed to commit to internal safety controls, independent external audits and independent board committees.
• The pact, announced on Sept. 29, is voluntary and described by President Trump as "morally binding," but it creates no new legal obligations for the companies that signed it.
• The agreement requires firms to set up internal security teams and outside auditors to monitor risks such as cyberattacks, biosafety threats, chemical hazards and models acting without authorization.
• Experts and analysts criticized the deal for concentrating governance power in a handful of U.S. firms, noting the timing ahead of planned IPOs and the absence of mandated public incident reporting or independent government oversight.
• The U.S. industry-led approach contrasts with the European Union’s binding AI Act and China’s state-led controls, increasing the likelihood of fragmented international rules and raising questions about transparency, enforcement and national sovereignty.