The US government has increased the list of AI companies favored by the administration through a deal signed by the Pentagon with Microsoft, Reflection AI (a company that hasn’t yet released its product to the public), Amazon, and Nvidia that allows them to use their products on classified missions. Apart from those named, OpenAI, xAI, and Google are among the firms that can be deployed “for any lawful use.”
“Any lawful use,” however, is the term that was central to the recent disagreement between Anthropic AI and the US government, which saw CEO Darius Amodei arguing that it would allow the US government to employ Anthropic’s technology in surveilling its civilians and developing autonomous weapons – uses of Anthropic’s technology that he wanted to keep out. Following his arguments, the Pentagon canceled a $200 million contract with Anthropic. Anthropic promptly sued the US government for hundreds of millions in losses due to the cancellation, losses not only to the US government but to other organizations influenced by the US government. It is noteworthy that during the Trump administration, Anthropic was labeled a “supply chain risk,” becoming the first company based in the US with such a designation.
Also Read : Physical AI Is Bringing Humanoid Robots to Real Factories
According to the Pentagon’s statement concerning the new deal, “The Department will continue to develop an architecture that mitigates vendor lock-in and provides for future flexibility for the Joint force.” In essence, the technologies are supposed to “provide warfighters the capabilities they need to operate confidently and ensure that our nation remains protected against any threat.” The AIs will be used in “Impact Levels Six (handling of classified information) and Seven (most sensitive classified information) applications,” creating an “AI-first fighting force.”
Currently, the use of generative AI by the Pentagon is limited to “non-classified tasks performed within defense departments including document writing and summarizing, and researching.” According to the statement, the new vendors will aid defense forces in the “streamlining of data synthesis” but also in “enhancement of situational understanding and warfighters’ decision-making in dynamic situations.” However, it is not stated whether these activities include operations conducted domestically.
The increase in the number of AI firms providing services to the US military and security agencies will make it difficult to be affected by any sudden changes in mind on the part of the individual firms that provide support to military and security operations. Through diversification, the decisions of individual executives in the companies will not matter much. Google and Amazon, in the past, have sacked employees who raised objections to the use of their firms’ technology in weapons and war.
The Claude AI from Anthropic has been deployed on classified information as part of Palantir’s Maven suite. But the latest signatories may take up this function. Nonetheless, the Mythos AI from Anthropic has been deployed at the National Security Agency of the United States in relation to its capabilities in cyber warfare and defence. Internationally, Anthropic’s Mythos is undergoing assessment by 40 entities out of which only 12 have been disclosed while MI5 of the United Kingdom and NSA of the United States could be among the other 28.
Also Read : RingCentral Expands AI Receptionist with Shopify, Calendly, and WhatsApp Integrations
Axios reports that the US government may be reversing its latest public position on Anthropic. Axios reported that it had a White House source who said the US government was looking for ways to save face and bring “’em back in.” The Claude language model from Anthropic is supposedly still being used by US government agencies concerned with national security, and that has been happening even during the latest events.
“[The US government] continues to proactively engage across government and industry to protect our country and the American people, including by working with frontier AI labs,” according to the White House.
