Welcome back to Daily Zaps, your regularly-scheduled dose of AI news ⚡

Here’s what we got for ya today:

Let’s get right into it!

Sign Up | Advertise | Tools

Anthropic has filed a federal lawsuit against the U.S. Department of Defense after the Pentagon designated the AI company a “supply-chain risk,” a move that could block it from government and military contracts worth hundreds of millions of dollars. The dispute stems from disagreements over whether the military should have unrestricted rights to use Anthropic’s AI technology, including for applications such as autonomous weapons.

Anthropic argues the designation is unlawful retaliation for its stance on limiting certain military uses and is asking the court to overturn the decision and allow it to continue selling to the government while the case proceeds. The Pentagon has broad authority to determine supply-chain risks, making the legal challenge difficult, but Anthropic may argue it was unfairly singled out, particularly after a competing company secured a new Pentagon contract. The case highlights growing tensions between AI developers and the U.S. government over how advanced AI systems can be used in military operations.

OpenAI has revised its agreement with the U.S. government on the use of its AI in classified military operations after criticism that the original deal was rushed and lacked clear safeguards. The updated terms explicitly prohibit using OpenAI’s systems for domestic surveillance of U.S. citizens and require additional approval before intelligence agencies like the NSA can use the technology.

The controversy emerged amid a broader dispute between the Pentagon and Anthropic, whose Claude model was labeled a supply-chain risk after the company resisted allowing its AI to be used for autonomous weapons or mass surveillance. The debate highlights growing tensions over how AI should be used in warfare, with companies, governments, and researchers grappling with issues such as military decision-making, human oversight, and the ethical limits of autonomous systems.

Anthropic has launched Claude Marketplace, a new platform that allows enterprise customers with existing Anthropic spending commitments to purchase third-party applications powered by Claude, such as tools from GitLab, Harvey, Replit, Snowflake, and others, with billing handled directly through Anthropic. The goal is to simplify procurement and consolidate AI spending while making it easier for companies to access specialized Claude-powered tools built by partners for specific industries and workflows.

The move signals that Anthropic sees continued value in SaaS applications enhanced with AI rather than fully replacing them with custom AI workflows. Claude Marketplace could allow enterprises to use Claude as a central orchestrator that connects to approved tools and services, though its success will depend on whether businesses adopt the marketplace instead of continuing to build their own AI integrations or using existing partner APIs.

Oracle and OpenAI have canceled plans to expand their large AI data center campus in Abilene, Texas after negotiations collapsed over financing terms and OpenAI’s shifting capacity needs. The site, part of the high profile Stargate AI infrastructure project announced at the White House, will still continue operating and developing its existing facilities, but the planned expansion from about 1.2 GW to roughly 2.0 GW of power capacity has been shelved.

The project has also faced local resistance and operational challenges, including cooling disruptions that temporarily shut down parts of the facility. As the expansion stalled, Crusoe, the site developer, began looking for new tenants, with Nvidia reportedly helping secure interest from Meta to occupy the unused capacity. Despite the setback, Oracle and OpenAI say their broader partnership to build large scale AI data center capacity across other locations remains in place.