As reported by Reuters, concerning the prior prohibition imposed by the US Department of Defense on the artificial intelligence startup Anthropic, the latest internal memo reveals that if the pertinent AI tools are deemed vital for national security, the Pentagon will permit specific units to retain and utilize Anthropic's products beyond the initially planned six-month phase-out period. This decision provides leeway for flexible exemptions in the implementation of the stringent ban. The memo, signed by Kirsten Davis, the Chief Information Officer of the Pentagon, on March 6, clearly states that exemptions will only be considered under rare and extraordinary circumstances, with stringent eligibility criteria. The AI tools must be employed for critical missions that directly bolster national security operations and lack viable alternatives. Furthermore, units seeking exemptions are required to present a comprehensive risk mitigation plan. Previously, Defense Secretary Pete Hegseth had identified Anthropic as a supply chain risk and implemented a total ban on its products for use by the Pentagon and its contractors. In reaction, Anthropic has initiated legal action to prevent the ban. Despite opening up the possibility for exemptions, the memo still mandates that priority be given to removing Anthropic's products from core critical systems. Pentagon contract officials have a 30-day window to inform contractors, who are then obligated to provide evidence of full compliance within 180 days.