(Credit: Ludovic Marin/Contributor/AFP via Getty Images)
Anthropic is having a huge moment, and it's getting under OpenAI CEO Sam Altman's skin.
Their feud runs deep, dating back to 2021, when Anthropic CEO Dario Amodei and his sister, President and Co-Founder Daniela Amodei, left OpenAI to found a more "trusted" company. The two have since built Anthropic into a runaway success, mostly thanks to a narrow focus on programming and Claude Code. Anthropic might even be getting a bit cocky; it aired a Super Bowl ad criticizing ChatGPT, prompting an X rant from Altman. The momentum has continued since then with the release of a new model, Sonnet 4.6, and a fresh $30 billion funding round—bringing its total funding past OpenAI's, Pitchbook reports.
The tension became visible this week when Altman—intentionally or not—seemed to avoid holding hands with Amodei on stage at the AI Summit 2026 in Delhi, India—despite the dozen other people in the lineup next to them who linked up without complaint.
Here's the backstory on the beef, and why Anthropic has much bigger fish to fry, including the US Secretary of Defense (War?) Pete Hegseth.
This Week's Big Story: Anthropic Is the Popular Kid in AI
Anthropic was the first, and perhaps the only, AI chatbot company I have reported on in the past few years that intentionally narrowed its focus. Whereas OpenAI spent its early days proving ChatGPT could do everything you need in your personal and professional life—Image generation! Videos! Voice mode! Essays! Science!—Anthropic went in the opposite direction. It decided to focus on the workplace, which I discussed with its Chief Product Officer Mike Krieger back in 2024.
The Claude chatbot cannot generate images, for example, which ChatGPT and Google Gemini consider table stakes for a fully functional chatbot. Instead, Anthropic focused on refining its B2B applications, pouring effort into products like Claude Code and, most recently, Claude Cowork. This helps Anthropic secure corporate contracts, making the case to businesses that they can save gobs of money on software engineers by using their product.
AI coding has taken off among the engineers, and has a much less controversial reputation than other AI-generated content. As one blogger aptly put it, "AI-generated code feels like progress and efficiency, while AI-generated articles and posts feel low-effort and make the dead internet theory harder to dismiss." Ironically, OpenAI debuted its coding tool, Codex, almost exactly at the same time as Anthropic released Claude Code in June 2025. But it hasn't taken off in the same way, so earlier this month, OpenAI released a Codex app for Mac to improve the user experience.
Meanwhile, Anthropic is fending off threats from another new foe: the White House, specifically the Department of Defense. Anthropic is suggesting limits on how the military uses its technology, reportedly insisting that Claude not be used for "the mass surveillance of Americans and fully autonomous weaponry." Secretary Hegseth is now threatening to cut ties with the company and, even worse, designate it as a federal "supply chain risk." That means any business that wants to work with the US military would have to cut ties with Anthropic.
Given Anthropic's momentum, it's unclear whether Hegseth's strong-arm tactics are timed well for success. Anthropic could use this moment to gain the moral high ground, a reputation Amodei has been eager to build through his media appearances and blog posts, and curry favor with the public (i.e., investors) for not capitulating.
Other Headlines You Need to Know
- The creator of OpenClaw joins OpenAI. The viral AI agent goes from a GitHub experiment to a corporate darling after Altman courts Pete Steiberger to join the ranks.
- Microsoft Copilot is reading confidential emails without consent. The company blames a code issue for Copilot rifling through emails specifically marked confidential, and says a fix is on the way.
- Micron invests $200 billion to ramp up memory production. As AI data centers hog the memory supply, Micron is pouring billions into an expansion at its Boise, Idaho, fab, months after it ditched the consumer business to focus on AI clients. Our expert Matt Buzzi has some tips on how to figure out how much RAM you really need amid the RAM crunch.
- Anthropic releases its latest model, Sonnet 4.6. It promises—you guessed it—improved coding skills, plus a massive context window, allowing users to upload entire codebases or hundreds of documents at once.
My Favorite Thing on the Internet Right Now
(Credit: Peon Ping)Calling all Warcraft III players: Someone built a plugin for Claude Code that plays commands from your computer you when your tasks are done—in the voice of a peon! Remember those green-faced workers who mined ore and built dwellings? They responded to your commands with phrases like:
- "Work, work"
- "Be happy to"
- "Something need doing?"
You can listen to all the sounds on the website for a mega dose of nostalgia, as well as a few from other computer games, like Starcraft.