Top Stories
Pentagon Used Claude AI in Iran Strikes - Hours After Trump Banned Anthropic
The US military used Anthropic’s Claude AI for weekend strikes on Iran, even as President Trump labeled the company a “supply chain risk” and ordered all federal agencies to stop using its products. According to CBS News, US Central Command systems used Claude to conduct intelligence assessments, identify targets, and simulate battlefield scenarios in real time.
The revelation underscores the chaotic nature of the Trump administration’s response to Anthropic’s refusal to remove guardrails around domestic surveillance and autonomous weapons. On Friday, Trump denounced Anthropic as a “Radical Left AI company” after the company refused Pentagon demands for unrestricted access. By Saturday, the military was actively using Claude in combat operations.
Anthropic CEO Dario Amodei told CBS News that the company had sought to draw “red lines” because “we believe that crossing those lines is contrary to American values, and we wanted to stand up for American values.” The Pentagon has not explained the contradiction between the official ban and the AI’s continued military deployment.
Source: CBS News
Tech Workers Launch Protests Against Military AI Use
The Anthropic controversy has sparked a broader reckoning across the tech industry. An open letter titled “We Will Not Be Divided” started with 236 Google employees and 65 OpenAI employees when it launched February 27, and reportedly grew to several hundred more in the following days.
At Google, employees are specifically demanding limits on how their AI work can be deployed by the military, echoing 2018 protests that forced Google to abandon its Project Maven Pentagon contract. The concerns intensified after reports emerged that Claude had been used in actual combat operations, making the ethical questions immediate rather than hypothetical.
The letter urges AI companies to “put aside their differences and stand together” to refuse Pentagon demands for “domestic mass surveillance and autonomously killing people without human oversight.” The wave of protest puts pressure on both AI companies and the Pentagon to establish clearer boundaries around military AI applications.
Sources: The Week, TechCrunch
Claude App Surges to #1 After Pentagon Ban
In an unexpected twist, Anthropic’s public battle with the Trump administration has proven to be remarkable marketing. Claude surged to the top of Apple’s App Store over the weekend, dethroning ChatGPT from the #1 spot it had held for months.
The surge began Saturday - the day after Trump’s ban announcement - and continued as news spread of the company’s defiance. Downloads increased even further when Health and Human Services announced it would begin phasing out Claude, leading to widespread media coverage of the standoff.
Consumer response suggests that Anthropic’s principled stance on surveillance and autonomous weapons has resonated with users increasingly concerned about AI ethics. The company’s willingness to lose government contracts rather than compromise on its values has created a stark contrast with OpenAI, which recently accepted Pentagon terms that Anthropic rejected.
Source: CNN
DeepSeek V4 Expected This Week
Chinese AI lab DeepSeek is expected to release V4 this week, timed to coincide with China’s “Two Sessions” parliamentary meetings beginning March 4. According to TechNode, the model will feature native multimodal capabilities - generating text, images, and video - and has been optimized for Huawei Ascend and Cambricon chips rather than Nvidia hardware.
The deliberate choice to build V4’s inference stack around domestic Chinese chips represents a significant shift in China’s semiconductor independence efforts despite US export controls. This would mark the first time a frontier AI model has been optimized primarily for non-Nvidia hardware.
Other reports suggest the model may be a trillion-parameter mixture of experts with a 1 million token context window, though DeepSeek has not officially confirmed these specifications. Benchmark claims circulating online have not been verified.
Source: TechNode
Quick Hits
-
HHS begins Claude phaseout: The Department of Health and Human Services has started removing Anthropic’s Claude from its systems following Trump’s executive order, though the transition will take weeks given how deeply integrated the AI has become in healthcare workflows. STAT News
-
Apple delays Gemini Siri features: Apple is reportedly pushing back some Gemini-powered Siri features beyond iOS 26.4, spreading capabilities across future releases through iOS 26.5 (May) and iOS 27 (September). According to 9to5Mac, features like personal data integration and voice-based in-app control are “running behind” due to snags in internal testing. TechCrunch
-
AI inference security concerns grow: Security researchers warn that while most AI discourse focuses on model training, the more immediate enterprise risk lies in inference - the operational phase where models are queried. Threats range from nation-state data extraction to unintended prompt leakage. The Quantum Insider
-
State privacy legislation heats up: Alabama’s HB 351 passed the House 103-0 and is now in a Senate committee with about a month until the legislature adjourns March 27. Connecticut’s SB 4, which includes facial recognition amendments and data broker provisions, entered committee February 26. Alaska’s HB 367 (Consumer Data Privacy Act) is in the House Judiciary Committee. The EU AI Act reaches full enforcement August 2.
-
London hosts largest anti-AI protest: Up to 500 protesters marched through London’s King’s Cross tech hub on February 28 in the “March Against the Machines.” The demonstration started outside OpenAI’s UK office, passed Google DeepMind and Meta headquarters, and ended with a People’s Assembly in Bloomsbury. Organizers demanded a pause on frontier AI development and democratic oversight. MIT Technology Review
Worth Watching
The Pentagon-Anthropic situation is evolving rapidly and raises fundamental questions about AI governance. How can a company be simultaneously banned as a “supply chain risk” and used for active military operations? The contradiction suggests either the ban is performative, or different parts of the government are operating without coordination.
DeepSeek V4’s release this week - timed to coincide with the “Two Sessions” parliamentary meetings - could mark a significant moment in the US-China AI race. The deliberate choice to optimize for Huawei Ascend chips rather than Nvidia hardware signals confidence in domestic semiconductor capabilities, though actual performance benchmarks will only be known after release.
The tech worker protests deserve attention too. Previous collective action at Google changed company policy on military AI. Whether this wave of protest achieves similar results will depend on how broadly it spreads and whether workers at other AI companies join.