Last Week in AI #337 - Anthropic Risk, QuitGPT, ChatGPT 5.4
Last Week in AI Last Week in AI
DOD reportedly flagged Anthropic as a supply-chain risk, while a 'cancel ChatGPT' movement is growing after OpenAI inked a deal with the US military. Two AI labs, two very different reputational headaches with the Pentagon.
Based on reporting by Last Week in AI, Last Week in AI — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
The Department of Defense has apparently put Anthropic on its radar as a supply chain risk, a designation that sounds bureaucratic until you remember it's the kind of label that can quietly kill contracts before they're even signed. Details on the exact reasoning are thin, but the label itself is the story: a company that built its entire brand on being the safety-conscious alternative to OpenAI is now getting side-eyed by the same government it's been courting for enterprise deals.
Meanwhile OpenAI has the opposite problem. After signing an agreement with the US military, a #QuitGPT-style backlash has been gaining steam online, with users and critics framing the deal as a betrayal of OpenAI's original nonprofit, humanity-first branding. It's not the first time OpenAI has taken heat for military ties, but the intensity this time suggests the goodwill Sam Altman's company has spent years building with everyday users isn't infinite.
Both stories point at the same tension. As frontier AI labs chase government contracts, largely because that's where the durable revenue and prestige is, they're running headfirst into two different constituencies who don't trust them for opposite reasons. The Pentagon wants assurances about security and reliability that Anthropic apparently hasn't fully delivered. The public wants assurances about ethics and restraint that OpenAI's critics say it's abandoning.
None of this happened in a vacuum. The show's hosts also flagged a busier-than-usual week of model releases, ChatGPT 5.4 among them, which barely got airtime compared to the drama around defense contracts. That imbalance says something: incremental model updates are becoming routine background noise, while who gets paid by whom, and who gets blamed for it, is where the actual news energy is going.
My take — AI-written commentary, not fact-checked reporting
I don't think either story is really about AI safety at all, it's about optics management for two companies that both desperately need government money and can't have it both ways. Anthropic wanted to be the trustworthy grown-up in the room and got flagged as a risk anyway; OpenAI wanted to keep its humanitarian glow while cashing defense checks and got called out for it. Pick a lane, guys.
Read more about this at: Last Week in AI
