Anthropic has announced plans to sue the Pentagon after the Department of Defense banned its software, calling the restriction unlawful. The legal threat marks a significant escalation between the AI company and the federal government, arriving at the same moment that broader tensions between Silicon Valley and Washington are playing out across multiple fronts.
CEO Dario Amodei has nonetheless issued an apology for a leaked internal memo critical of the Trump administration. The president, for his part, said he fired Anthropic “like dogs,” a remark that drew wide attention. Separately, Anthropic received a measure of relief when its models were confirmed to remain available within Microsoft products.
The Pentagon’s Long History With AI Models
The Anthropic dispute arrives alongside a separate revelation: the Pentagon has been quietly testing OpenAI models for years. The disclosure raises pointed questions about how meaningful OpenAI’s earlier ban on military use of its models ever was in practice. The gap between stated policy and operational reality appears to have been substantial.
Dean Ball, a former Trump administration AI adviser, captured the contradiction bluntly. “It’s sort of bitterly ironic,” he told Politico, pointing out that the Anthropic dispute runs directly counter to the president’s stated commitment to cutting bureaucratic red tape for the technology sector.
Iran Strikes Amazon Data Centers
In what is being described as the first military strike on a US hyperscaler, Iran hit Amazon data centers in the Gulf region, rattling the area’s technology sector. The attack has drawn fresh scrutiny to AI’s current and projected role in modern warfare, with researchers and policymakers now examining where that trajectory leads.
The incident carries implications well beyond the immediate damage. Gulf nations have invested heavily in positioning themselves as AI infrastructure hubs, and a direct hit on American cloud assets tests both the physical security assumptions behind that strategy and the geopolitical calculations driving it.
AI and Energy: A Pledge Without Hard Numbers
On the domestic side, Google, Microsoft, Meta, Amazon, OpenAI, Oracle and xAI have all signed a pledge alongside Trump and tech executives to protect consumers from AI’s energy costs. The commitment has been met with skepticism in some quarters, particularly given the difficulty of independently verifying AI’s true energy footprint.
Other Developments Worth Tracking
- A new lawsuit argues that Trump’s TikTok deal benefited firms that “personally enriched” him, seeking to reverse the sale of the app’s US operations.
- Meta faces a lawsuit alleging it misled users about the privacy features of its smart glasses, with the suit claiming the devices enabled covert surveillance.
- Researchers have established a new field studying “AI societies,” using AI agents to examine human behavioral patterns without involving human participants at all.
- Google and Amazon are pushing smarter home assistants as a potential reboot for the stalled smart home market, though consumer enthusiasm remains mixed.
The US National Video Game Museum has acquired the development kit for the mythical Nintendo PlayStation, the fabled hybrid console prototype that was never commercially released, closing a long chapter of gaming history speculation.
Meanwhile, MIT Technology Review’s AI team is preparing a special report, “10 Things That Matter in AI Right Now,” set to publish in April alongside its EmTech AI event. The report will feature perspectives from leaders at OpenAI, Walmart, General Motors and others, spanning AI agents, business infrastructure, and the future of human expression.
Photo by Saifee Art on Unsplash
This article is a curated summary based on third-party sources. Source: Read the original article
