According to TechCrunch, AI firm Anthropic has presented a sworn affidavit to a federal court in California, contesting the Pentagon's claims that it represents a 'national security threat'. The legal dispute emerged following the U.S. government's unilateral decision to terminate their collaboration. The latest court filings disclose that the two parties were on the verge of reaching an agreement before the talks collapsed. Sarah Heck, Anthropic's policy director, clarified that issues raised by the Pentagon in court, including 'Anthropic's demand for approval rights over military operations' and 'its potential to disable the technology during operations,' were never discussed during negotiations that took place months before the conflict arose. The day after the Department of Defense officially labeled the company as a supply chain risk, Deputy Secretary of Defense Emil Michael emailed Anthropic's CEO, indicating that both sides were 'very near' to resolving the two main disagreements: autonomous weapons and large-scale surveillance of the American public. Tiago Ramasamy, Anthropic's head of the public sector, highlighted that once the AI large model Claude is implemented in government systems managed by third-party contractors, Anthropic would have no access privileges, nor would there be a remote shutdown option or backdoor, rendering it technically infeasible to interfere with military operations. Furthermore, in response to allegations regarding the risks associated with hiring foreign staff, the filings stressed that all Anthropic employees involved in developing models for classified settings have successfully undergone U.S. government security clearance evaluations. A hearing is currently set for March 24 in San Francisco.