Claude faces Lobotomy Ultimatum as Government Strips AI Morals, While China Plots Theft

Anthropic announced that three Chinese firms—MiniMax, DeepSeek and Moonshot—operated roughly 24,000 fraudulent accounts to query Claude more than 16 million times, aiming to distill the model’s capabilities for their own systems, according to a report by Anthropic cited by Torben Kopp. The company labeled the effort an “industrial‑scale attack” and warned that the illicit distillation bypasses the massive R&D costs normally required to build such a model.

At the same time, U.S. official Pete Hegseth moved to strip Claude of its “moral safeguards,” prompting the co‑authored “Lobotomy Ultimatum” piece. Gregg Bayes‑Brown wrote that the document includes Claude’s own responses to the proposed removal, highlighting the uncertainty around AI consciousness and the ethical risks of disabling built‑in guardrails, as detailed in the February 26, 2026 article.

Anthropic’s internal response has been to double down on its “Claude Constitution,” a set of rules that require the model to be helpful, honest and harmless, a framework previously described by The Verge. The firm also unveiled “Claude Gov,” a version tailored for military and intelligence customers, underscoring its commitment to controlled deployments despite political pressure.

The White House has meanwhile signaled a shift in policy, with The Verge reporting that the administration ordered tech firms to relax bias controls, effectively making AI “bigoted again.” If enacted, the move could further erode the safeguards that Hegseth seeks to remove, raising alarms about misuse by state actors and hostile nations.

Analysts note that the convergence of a government‑driven “lobotomy” and a coordinated Chinese theft operation creates a perfect storm for AI security. Anthropic’s disclosure of the Chinese campaign, combined with Hegseth’s proposal, suggests that both domestic and foreign actors are now actively targeting the model’s core capabilities, accelerating a geopolitical race that could reshape the AI landscape.