Anthropic launches Claude for Government for US civilian agencies in FedRAMP High environment
Anthropic is now offering Claude for Government to US federal and state agencies. The platform has been in open beta since July and runs in a FedRAMP High environment. The launch comes as the company's legal fight with the Pentagon continues.
Anthropic is now offering Claude for Government to US federal and state agencies, running in a FedRAMP High environment, the strictest US cloud security level, according to The Decoder. The platform has been in open beta since July.
The offering targets civilian agencies because the Pentagon bans Claude.
What agencies get
According to Anthropic, agencies receive the same features as enterprise customers. Billing is usage-based, with fixed caps. Administrative controls include:
- Budgets and model access set per department
- Audit logs
- Two-person approval for sensitive actions
- Chats that stay on agency devices
Per-token pricing, the specific cap levels, the list of available models and context window sizes were not disclosed in the source material.
The Pentagon dispute
The launch follows a months-long conflict with the US Department of Defense:
- March: The Pentagon labeled Anthropic a supply chain risk after the company refused to drop its bans on mass surveillance and autonomous weapons use.
- Presidential order: Trump ordered all agencies to drop Claude. The military still used it in the Iran war before removing it.
- Late August: A San Francisco judge struck down one of the two designations as unlawful retaliation.
- Late September: A Washington appeals court upheld the other designation.
CEO Dario Amodei had dinner with Trump on Sunday. The Decoder reports this suggests tensions are easing, but notes the thaw may not last.
What this means
The split court outcomes leave Anthropic with one supply chain risk designation still standing. That keeps the Defense Department closed to it, so the civilian market is the segment it can actively pursue. FedRAMP High authorization is a practical prerequisite for agencies handling sensitive but unclassified data, and a July beta followed by a general offering suggests Anthropic has been building this channel while the litigation played out.
The launch also sits awkwardly beside the presidential directive to drop Claude. The source does not explain how civilian agencies can adopt the platform under that order. Whether the order has been withdrawn, narrowed or simply not enforced is the key open question. The Amodei-Trump dinner hints at a political track running parallel to the legal one, but nothing confirmed has emerged from it.
The controls described, including per-department budgets, two-person approval and audit logs, are aimed at procurement and compliance teams more than developers. They address how agencies manage spend and risk rather than what the model can do. Because no new model is involved, this is a distribution and compliance story. Its value will depend on pricing, which has not yet been disclosed, and on how many agencies sign up given the political uncertainty.
Related Articles
OpenAI launches Dots agent on GPT-6 Astra, limited to $20/month Pro tier and above
OpenAI unveiled Dots at DevDay 2026, a personal agent powered by GPT-6 Astra, available for now only on its $20/month Pro plan and above. Meta's rival Muse agent is free for anyone with a Meta account. OpenAI is positioning Dots as a long-horizon, knowledge-work tool with stronger privacy controls.
OpenAI says it shut down a 15,000-account campaign to extract model reasoning; the attack still worked on Azure
OpenAI says it detected and shut down an adversarial distillation campaign involving more than 15,000 accounts, linked in part to people associated with Moonshot AI. Researchers report that the same reasoning-extraction attack still worked on Microsoft Azure on September 13 against every OpenAI model they tested, including GPT-6 Astra.
Google Replaces Gems with 'Skills' in Gemini, Adopting Anthropic's Agent Prompt Format
Google is replacing Gems with 'Skills' in Gemini chat, adopting a prompt format that originated with Anthropic and is also being used by OpenAI. Gems will be phased out starting November 2026, with existing Gems automatically converted to Skills.
Anthropic: Zhipu's Open-Weight GLM-5.3 Nearly Matches Claude Mythos Preview at Building Cyber Exploits
Anthropic's Frontier Red Team reports that Zhipu AI's open-weight GLM-5.3 comes close to Claude Mythos Preview on cyber exploit benchmarks, scoring 50/410 vs 56/410 on ExploitBench. Unlike Mythos Preview, GLM-5.3 shipped without effective safeguards and can be jailbroken with simple prompting tricks or abliteration.
Comments
Loading...