OpenAI and Anthropic Push for Independent AI Safety Testing Amid Debate Over Government Authority

OpenAI and Anthropic are advocating for independent pre-deployment testing of advanced AI systems, raising concerns about potential risks to humanity. While both companies work with existing federal evaluator CAISI and independent organizations, experts note there are no universal testing standards. The push coincides with investor interest and upcoming stock market listings, prompting questions about whether commercial motivations are influencing the safety narrative. Political figures including President Trump and advisors have dismissed calls for regulation, while former employees and researchers continue to raise concerns about AI risks including bioweapons and nuclear escalation.

Her position was that this redirection of attention away from present-day harms deserves scrutiny, citing the use of artificial intelligence within weapons systems that she maintained have already been linked to deaths. The conversation drew renewed attention this week after Anthropic engineer Jacob Coxon resigned, calling for a pause on the development of advanced AI systems. OpenAI spokesperson Liz Bourgeois stated that the company had paused work on its most advanced AI models, adding that trust in responsible AI development rests on the conduct of industry participants, while President Donald Trump has described concerns about AI risks as manufactured to benefit China and has taken a position against new regulatory measures in the field. David Sacks, a venture capitalist who serves as co-chair of Trump's Council of Advisors on Science and Technology, has labelled those who favour a more measured approach a "Doomer Industrial Complex." A federal body tasked with assessing artificial intelligence already operates within the country, functioning under the name CAISI — a restructured iteration of the AI Safety Institute that was established under the Biden administration in 2023 and given a new designation in 2025. PitchBook analyst Harrison Rolfes observed that the arrangement could serve a commercial purpose, placing both firms in a stronger position with investors and partners such as Nvidia and Google, and that the two organisations appeared to be establishing a competitive advantage within the industry, with all parties set to gain financially — while Nvidia chief executive Jensen Huang told Trump separately that he viewed concerns about artificial intelligence as overstated.

OpenAI and Anthropic Push for Independent AI Safety Testing Amid Debate Over Government Authority

Related Posts

OpenAI and Anthropic Push for Independent AI Safety Testing Amid Debate Over Government Authority

Related Posts