"The Pentagon is also peeved by Anthropic dictating the technology's terms of use: The company wants assurances that its tool won't be used to direct autonomous weapons or surveil Americans. The two sides signed a deal in July worth up to $200 million, but clearly didn't hash out their potential differences."
"According to reports that cite Pentagon officials, the Donald Trump appointee gave Anthropic an ultimatum on Tuesday: If the company doesn't relax its AI's safeguards, the Pentagon will label Anthropic a 'supply chain risk' or invoke the Defense Production Act to push the company into compliance."
"My sense is that the administration is trying to box in Anthropic, by on the one hand threatening to really harm its business by labeling it a 'supply chain risk,' and on the other saying, 'We're going to invoke the DPA, and if you don't like that, you're going to have to go through the' - Joel Dodge, policy expert at Vanderbilt Law School"
Anthropic, an AI safety-focused startup, faces pressure from Defense Secretary Pete Hegseth to relax safeguards on its AI technology used for classified military work. The Pentagon wants fewer restrictions on the company's cutting-edge AI model, which was reportedly used in capturing Venezuelan leader Nicolas Maduro. Anthropic has insisted on assurances preventing autonomous weapons deployment and American surveillance. The conflict stems from a July deal worth up to $200 million that apparently left key differences unresolved. Hegseth has threatened to label Anthropic a supply chain risk or invoke the Defense Production Act to force compliance, creating contradictory pressures that simultaneously threaten business harm and forced cooperation.
Read at SFGATE
Unable to calculate read time
Collection
[
|
...
]