Sen. Warren questions DOD about Anthropic blacklist that 'appears to be retaliation'

Senator Elizabeth Warren, representing Massachusetts, has raised serious concerns regarding the Department of Defense's (DOD) classification of the artificial intelligence startup Anthropic as a supply chain risk, suggesting that this move may be an act of retaliation. In a letter addressed to Defense Secretary Pete Hegseth, Warren pointed out that the DOD had alternative options, such as terminating its contract with Anthropic or continuing to utilize its technology in unclassified environments. Warren expressed her apprehensions that the DOD may be coercing American companies into providing tools that could infringe on the privacy of citizens, as well as enabling the development of fully autonomous weapons lacking sufficient safeguards. As the conflict in Iran progresses into its fourth week, U.S. senators are demanding more transparency regarding the DOD's collaborations with technology firms. Tensions escalated between the DOD and Anthropic prior to the outbreak of hostilities, as the department sought unrestricted access to Anthropic’s models for various lawful applications. In contrast, Anthropic insisted on guarantees that its models would not be employed for autonomous weaponry or domestic surveillance. On February 27, Secretary Hegseth announced the decision to officially label Anthropic with the “supply chain risk” designation. This announcement came just as the DOD continued to utilize Anthropic's AI model, Claude, in operations related to the ongoing conflict in Iran. Following the blacklisting, Anthropic initiated legal action against the Trump administration, claiming the designation posed a threat to national security. A preliminary hearing for this case is set to take place on Tuesday in the U.S. District Court for the Northern District of California. In the aftermath of Anthropic's blacklisting, OpenAI swiftly stepped in, announcing a partnership with the DOD. OpenAI expressed confidence that its AI solutions would not be utilized for mass surveillance or the creation of fully autonomous weapons, citing its safety protocols and existing legal frameworks, although the complete contract details remain undisclosed. Nonetheless, concerns persist among lawmakers, the public, and some OpenAI employees. Warren is also seeking clarity from OpenAI's CEO, Sam Altman, regarding the specifics of the agreement with the DOD. In her letter, she voiced her worries that the terms might allow the Trump administration to exploit OpenAI’s technology for invasive surveillance of Americans and to develop lethal autonomous systems that could pose risks to civilians with minimal human oversight. During a recent meeting in Washington D.C., Altman faced tough questions from lawmakers, including Senator Mark Kelly from Arizona, who expressed significant doubts about OpenAI's strategies related to warfare and its DOD contract. Warren emphasized the necessity of reviewing the complete agreement to evaluate the safeguards and restrictions it may include, a request that remains unanswered as both the DOD and OpenAI have not released the full contract. She concluded that the limited disclosures raise alarming questions about the DOD's AI applications. With the Democratic Party in the Senate facing challenges to instigate action, the situation remains precarious as the Republicans maintain control over the White House and Congress.