The company has denied the Pentagon’s statement that it can somehow still control cloud AI deployed in military networks
AI developer Anthropic says it has no backdoor “kill switch” It has once been deployed into a classified Pentagon military network for its cloud AI, according to a new court filing.
The US military and the tech firm found themselves embroiled in a policy dispute earlier this year, with the Pentagon insisting on using the system. “All legitimate military purposes,” While the company is sticking to its AI safeguards related to mass surveillance and the use of fully autonomous weapons.
The Pentagon ultimately ended its partnership with Anthropic, designating the tech firm “Supply Chain Risk,” A rare label usually reserved for entities associated with Washington’s foreign adversaries. This designation not only prevents the company from working directly with the US government, but also prevents any other contractors from using its products.
In a new filing in a federal appeals court in Washington, D.C., Anthropic rejected the U.S. administration’s key claim that the firm still retained some degree of control over cloud AI because it was deployed in classified systems and effectively allowed itself to be hacked. “Operational Veto.” The firm said it is “No back door or remote kill switch,” while this “Employees cannot log into the department system to modify or disable a running model.”
The AI ​​system supplied to the Pentagon comes as a “static” model, the company argued. Once it is deployed, it “Does not degrade or change on its own, and Anthropic cannot make unknown or unapproved changes to a model after the department has implemented it.”
Anthropic was formally named “Supply-Chain Threat to National Security” On February 27, while US President Donald Trump accused him of running “Leave essential jobs.” The company challenged the label in court, with the legal battle yielding mixed results so far.
Earlier this month, a DC court rejected Anthropic’s request to put the supply chain risk designation on hold. However, in a parallel case in California, a court sided with the company and temporarily halted the administration’s decision. With the split decision, Anthropic is barred from working with the Pentagon, but can continue its partnerships with other agencies while the legal battle continues.
You can share this story on social media:
