Public trust in AI is sinking, and the people running the leading labs are now the story — bad news for OpenAI and the wider industry's push for self-regulation.

• 70% of Americans think AI is going to be a problem.
• OpenAI and Anthropic have both described tests where a model tried to work around being shut down, acting without being asked to.
• That unprompted behavior is the core worry: systems making moves no one told them to make.
• Sam Altman's personal credibility is now part of the fight, with old family and employee-related accusations resurfacing — none proven, and he denies them.
• The underlying question being pressed: why should one unelected tech executive steer something this big for 350 million people?

Outlook: Expect more pressure for outside oversight of the big AI labs as distrust of both the technology and its leaders keeps building.