Trump Cuts Federal Ties With Anthropic AI Over Security Fears

Trump just ordered federal agencies to dump Anthropic’s AI products within six months. The move comes after Pentagon brass raised red flags about the company’s technology and whether it’s secure enough for military work.

The whole mess started when defense officials got worried that Anthropic’s systems might have security holes that enemies could exploit. Pentagon sources said the vulnerabilities were serious enough to make military leaders nervous about keeping the tech around sensitive operations. And once those concerns hit Trump’s desk, he didn’t waste time making the call. Federal agencies now have to scramble to find new AI providers that can meet tougher security standards. The timeline’s pretty tight, but the administration wants this done fast.

Not exactly great news for Anthropic.

The company hasn’t said much publicly about Trump’s directive yet, but you can bet they’re feeling the heat. Anthropic’s AI tools have been spread across tons of government functions, so pulling them out won’t be simple. Agencies are looking at a massive logistical headache as they try to replace systems that have been running their operations. Some departments already started shopping around for alternatives, but the six-month deadline means they can’t take their time.

Defense Secretary Mark Esper said his team’s working with tech consultants to find AI providers with better security features. The Pentagon wants systems that can handle classified work without creating backdoors for hackers or foreign governments. Esper’s people are basically starting from scratch, evaluating new companies and making sure their tech can pass military-grade security tests.

The White House hasn’t spelled out exactly what agencies need to look for in replacement technologies. That’s left departments guessing about what standards they need to hit. Related coverage: Trump Bans Anthropic Tech from All.

Federal workers are already reporting headaches trying to figure out the transition. The General Services Administration jumped in to help agencies find new vendors and navigate the procurement process. GSA officials know the clock’s ticking, so they’re fast-tracking approvals and cutting through red tape where they can. But even with that help, some agencies are struggling to move fast enough.

A White House official who didn’t want to be named said the administration’s ready to throw money at the problem if needed. The government will help cover transition costs to make sure agencies can meet the deadline without breaking their budgets. That shows how serious Trump is about getting Anthropic’s tech out of federal systems, even if it costs taxpayers extra cash.

The Department of Homeland Security is probably feeling the biggest pinch right now. DHS has been using Anthropic’s AI for data analysis work, and pulling it out could mess up ongoing investigations and security operations. DHS spokesperson Emily Davis said on February 27 that her department’s actively hunting for new vendors to keep things running smoothly. But finding replacements that can handle the same workload isn’t going to be easy.

Wall Street’s paying attention too. Goldman Sachs analysts think federal contracts make up about 15% of Anthropic’s yearly revenue, so losing that business is going to hurt. The company’s stock took a hit when news of Trump’s order broke, and investors are wondering how Anthropic plans to make up for the lost income. CEO Dario Amodei has an internal meeting scheduled for February 28 where he’s expected to lay out the company’s game plan for dealing with the federal ban. This follows earlier reporting on Block Slashes 40% of Workforce in.

Politicians on Capitol Hill are split on whether Trump made the right call. Senator Tom Cotton from the Armed Services Committee backed the decision, saying the government can’t mess around when it comes to AI security. But some Democrats think the administration should have taken a slower approach instead of forcing agencies to make such a quick switch. They’re worried that rushing the transition could create new security problems or leave gaps in government operations.

Other AI companies are basically licking their chops right now. OpenAI and Google DeepMind are both positioning themselves to grab the federal contracts that Anthropic’s about to lose. Both companies have been talking up their security credentials and telling government buyers they’re ready to step in. The competition for those contracts is going to be fierce, especially since federal work tends to be pretty lucrative and stable.

Agencies have their work cut out for them over the next few months. They need to find new AI providers, test their systems, train staff on new tools, and make sure everything’s working before the deadline hits. Some departments are already behind schedule, which means they’ll be working overtime to avoid getting caught with their pants down when the six months are up.

The whole situation’s still pretty fluid, with more changes probably coming as agencies figure out their next moves.