Anthropic has been drawing heavy criticism after reportedly supporting legislation that would allow an AI kill switch to be used only if a system caused more than 50 human deaths or serious injuries or more than $1 billion in damage.
Here's what to know
Based on a POLITICO report, More Perfect Union (@PerfectUnion) said in an Instagram post that Anthropic has backed AI safety legislation that allows intervention only when a model creates an "imminent, enumerated risk."
According to More Perfect Union, that standard would cover harms tied only to a biological weapon, a cyberattack, or a loss of developer control. Further, it would apply only once the toll exceeded 50 deaths or serious injuries or $1 billion in damages.
Critics were quick to point out that the high threshold appeared to be at odds with past statements from Dario Amadei, Anthropic's CEO.
"I would rather they make fun of me than to wake up one day and discover that someone used our model, Claude, to kill a bunch of people," Amadei says in More Perfect Union's Instagram video.
The video also flags what it deems "preemption language," which would mean that the federal bill, if passed, would supersede state-level AI laws. If that language were adopted, states could be prevented from enforcing laws that "suspend the development, deployment, or internal use of an AI model," potentially undercutting stricter state rules, according to More Perfect Union.
TCD Picks » Quince Spotlight
💡These best-sellers from Quince deliver affordable, sustainable luxury for all
Commenters were quick to voice their skepticism.
"Fifty people ever or 50 at once?" wrote one. "I think we know the answer."
"Should be zero," added another.
More background
Another flashpoint for the controversial legislation has been the limited set of harms that would count toward the cutoff threshold. According to More Perfect Union, only biological weapons, cyberattacks, and cases where a model slipped out of developer control would qualify.
In the comments, some Instagram users focused on who should be held accountable in instances where AI has caused harm, with some arguing that the AI companies should be responsible while others pushed for operator liability.
Rather than waiting for mass casualties or billion-dollar damage, critics argued for rules that place clearer responsibility on the people and companies deploying these systems.
"The person who owns and operates the AI should be responsible as if they did it themselves," said one commenter. "What is the difficulty with that?"
Where can I learn more?
The fight over AI oversight has been showing up in Congress, statehouses, and beyond.
• Across the industry, AI risks have jumped, intensifying calls for tougher oversight.
• In Congress, AI's climate and public health toll has become harder for lawmakers to dismiss.
• In California, lawmakers wanted data center water transparency as AI expansion strained local planning.
• Public anger has grown as AI strains jobs, water, and electricity supplies.
• At the UN climate summit, AI's 'positive force' status faced growing skepticism.
Get TCD's free newsletters for easy tips, smart advice, and a chance to earn $5,000 toward home upgrades. To see more stories like this one, change your Google preferences here.
