The US Government Just Dumped Anthropic for OpenAI — Here's What Happened

# The US Government Just Dumped Anthropic for OpenAI — Here's What Happened

The Trump administration just delivered a massive blow to Anthropic. In the span of 72 hours, multiple federal agencies have terminated or are phasing out their use of Claude — and OpenAI is swooping in to pick up the contracts.

On February 27, President Trump directed all U.S. government agencies to cease use of Anthropic technology, declaring it a supply-chain risk. Since then:

Pentagon: First to announce a phase-out, then signed a deal with OpenAI to deploy AI on classified military networks
Treasury: Secretary Scott Bessent announced the department is terminating all use of Anthropic products, including Claude
State Department: Switching from Anthropic's Claude to OpenAI's GPT-4.1 for its in-house chatbot, StateChat
Health and Human Services (HHS): Notified employees to stop using Claude and switch to ChatGPT or Gemini
Federal Housing Finance Agency: Terminated all Anthropic products across Fannie Mae and Freddie Mac

That's four cabinet-level agencies plus housing regulators — all walking away from Claude in less than a week.

The standoff wasn't about performance or pricing. It was about what AI is allowed to do:

Autonomous weapons: The Pentagon wanted Claude capabilities for AI systems that could kill without human oversight. Anthropic drew a hard line.

Domestic surveillance: The administration wanted guarantees Claude wouldn't be used to spy on Americans. Anthropic refused to give blanket assurances.

Who controls deployment: Anthropic wanted guardrails the government couldn't override. The government wanted full control.

OpenAI, by contrast, agreed to deploy on classified military networks. CEO Sam Altman said they'll "amend" the Pentagon deal to clarify AI won't be used for domestic surveillance — but the core deal stands.

This isn't just political — it's $60 billion at stake. Anthropic's partnerships with Google and Amazon were supposed to be its government on-ramp. Now those same agencies are being told to dump anything connected to Anthropic's backers.

The State Department's switch is particularly notable: they specifically cited GPT-4.1 as the replacement. That's a win for OpenAI's newest model.

OpenAI dominates government: If you're doing AI business with the US federal government, you're probably going through Microsoft and OpenAI now
Anthropic's enterprise play takes a hit: Losing federal contracts hurts credibility with enterprise customers
The safety vs. access debate is over: The government chose access. Anthropic's safety-first stance cost them
Expect more churn: If your company uses Claude, expect procurement questions soon

Anthropic drew a line on surveillance and autonomous weapons. The government called their bluff and walked. OpenAI is now the default AI partner for the US federal government — and they're making sure everyone knows it.

This is the biggest AI politics story of the year so far. Watch for more agencies to make the switch in the coming weeks.

What Happened