Anthropic Got Shown The Door, And OpenAI Already Has The Keys

Nobody told us 2026 would feel like a workplace drama set inside a data centre. Yet here we are — watching governments fire chatbots, and chatbots take over computers, all within the same news cycle.

A Disagreement About Paperwork That Ended a Defence Contract

Let’s start with the messier story. The Pentagon has officially stamped Anthropic — the company behind the Claude AI — as a “supply chain risk,” effective immediately. For those unfamiliar with defence-contractor bureaucracy, that’s roughly the equivalent of being not just fired, but walked out by security while your colleagues are warned not to make eye contact.

What makes this particularly eyebrow-raising is that this designation has historically been reserved for foreign adversaries — think Huawei, think state-linked firms with alleged ties to Beijing. Anthropic is a startup from San Francisco.

Also Read: Motorola Edge 70 Fusion launched in India with Sony LYTIA 710 Camera and 7000 mAh battery

So what was the great sin? Anthropic pushed for guardrails that would explicitly ban the US military from using its Claude model to conduct mass surveillance on Americans or power fully autonomous weapons. The Pentagon, for its part, insisted it needed Claude for “all lawful purposes,” and argued that the uses Anthropic feared were already not allowed. In short: a disagreement about paperwork ended a defence contract. Classic.

Trump called the company “radical left” and “woke,” while Hegseth labeled Anthropic “sanctimonious” and another official said CEO Dario Amodei had a “God complex.” Amodei, for his part, reportedly told staff that part of the problem was that his company had not given “dictator-style praise to Trump” — which, when you think about it, is a fairly unusual line item in a government contract negotiation.

Also Read: OPPO Find X9 Series to Get AirDrop Compatibility via Quick Share Update

Your New Digital Coworker: GPT-5.4

In the midst of the dispute, OpenAI forged its own deal with the Department to allow the military to use its AI systems for “all lawful purposes.” Convenient timing. While Anthropic was getting escorted out the door, its biggest competitor quietly slipped in through the back.

And OpenAI had more good news to announce beyond just inheriting Anthropic’s government client. Meet GPT-5.4 — and before you groan at another model number, this one actually does something new. Rather than sitting in a chat window waiting to be asked things, it can now take over your computer and just… do stuff. Open Excel, draft a report, click through a presentation — all without you lifting a finger. Less “helpful assistant,” more “colleague who doesn’t wait to be asked twice.”

Also Read: Samsung Galaxy S26 Series Uses 8-Bit Panels, Not 10-Bit

It clicks, it types, it opens applications, and presumably wonders why you have 47 browser tabs open. Internal testing shows that statements generated by the model are about 33 percent less likely to contain false information compared with GPT-5.2 — which, if you’ve spent any time arguing with a chatbot about basic facts, feels like meaningful progress.

Taken together, these two stories paint a vivid portrait of where AI stands right now: one company is being blacklisted for trying to stop its AI from going rogue, while another is launching an AI specifically designed to operate your computer autonomously. The future is here, and it’s complicated, litigious, and apparently running on a six-month offboarding schedule.

Also Read: Oppo’s Best Camera Phone Is Finally Leaving China: Here’s What We Know So Far

The big takeaway? If you want to keep working with the US government, don’t ask too many questions. And maybe let the AI handle your emails while you figure out the rest.

You can follow Smartprix on Twitter, Facebook, Instagram, and Google News. Visit smartprix.com for the latest tech and auto news, reviews, and guides.