AI CEOs urge government oversight of advanced models

Anthropic and OpenAI CEOs have declared their advanced AI models dangerous and are calling for government regulation and independent testing before release, a rare show of unity from the rival companies. The push comes as the firms seek to shape safety protocols in an unregulated market, potentially influencing political goals ahead of US midterm elections and attracting investors before anticipated public stock listings. Critics argue the focus on existential risks diverts attention from current harms like AI-enabled surveillance and military use. President Donald Trump has dismissed AI risk warnings as a hoax, and his administration has not pursued new regulations. The companies are instead proposing their own auditing parameters and selecting evaluators, a move analysts say could create a competitive moat that benefits the largest labs financially.