Sanders vs. AI Giants: Safety Pledges Quietly Dropped

Senator Bernie Sanders has issued a strong ultimatum to the chief executives of OpenAI, Anthropic, and Meta. He is demanding they voluntarily pause their most powerful AI development or face legislative action. This move was coordinated with 29 House Democrats sending similar inquiries. Here's the thing: Sanders' letter invokes specific safety pledges he says these companies made. However, an independent safety audit documented over a month ago that the companies had already quietly retracted these very pledges. For example, Anthropic dropped its pledge in February 2026 to guarantee safety measures before training a system. The Future of Life Institute's Summer 2026 AI Safety Index, published July 7, found that Anthropic, OpenAI, Google DeepMind, and Meta had all weakened or voided their prior commitments to pause development if systems became too dangerous. The panel called this "moving the goalposts." The bottom line is that the words Sanders is using as his legal basis may no longer represent the companies' current commitments. This situation highlights the evolving and complex nature of AI safety regulations.