The Week the AI Industry Picked Sides
Week of May 11 to 18, 2026 | Issue 6 | For smart professionals catching up on AI
This week
Three years of "AI for everyone" ended this week. OpenAI is preparing to sue Apple over a partnership that was supposed to put ChatGPT inside Siri. Anthropic refused a Chinese think tank access to its most powerful model and the same week confirmed the Pentagon is now using it. Google rolled "Gemini Intelligence" across Android, Chromebooks, cars, and your living room. And a new startup called Recursive Superintelligence raised $650 million to build AI that improves itself. The companies that pretended to be neutral utilities have stopped pretending. If you use one of these tools at work, your choice of vendor is becoming a choice of side.
🗞️ News you should know
1. OpenAI is preparing to sue Apple over ChatGPT in Siri
OpenAI has hired outside lawyers to draft a breach-of-contract notice against Apple, according to Bloomberg's Mark Gurman on May 14. The two-year-old deal that put ChatGPT inside iOS 18's Siri never delivered the paid-subscription growth OpenAI had projected. Apple stock dipped 1% on the news. Apple, separately, is reportedly testing Google's Gemini as the new brain for a revamped Siri due in 2027, which would push OpenAI off the iPhone entirely. What this means for you: if you bought an iPhone in the last year because the marketing promised "ChatGPT inside Siri," that integration is the most likely casualty of this fight. Stop relying on it. Use the standalone ChatGPT app or the standalone Claude app for anything that matters. They both work fine on iOS, and neither depends on Apple staying friends with anyone.
Source: arstechnica.com | Published: 2026-05-14 Corroboration: 9to5mac.com
2. The Pentagon is now running Anthropic's Mythos despite calling the company a supply-chain risk
Reuters confirmed on May 12 that the Department of Defense is deploying Anthropic's Mythos model to find and patch software vulnerabilities across U.S. military networks. This is the same Anthropic the Pentagon labeled a supply-chain risk in March, and the same Mythos that Anthropic refused to give a Chinese think tank access to one week earlier. The White House National Security Council reportedly pushed for that refusal. Anthropic's revenue grew 80x year over year in Q1 and its annualized run rate now exceeds $44 billion, making it the first AI company to overtake OpenAI in verified business customers. What this means for you: if your company has a security-sensitive workload and your IT team is choosing between Claude and ChatGPT, the gap widened this week. Anthropic spent six months convincing Washington it was the trustworthy AI lab, and it appears to have worked. Expect Claude features to start showing up in your enterprise security and compliance tooling first.
Source: reuters.com | Published: 2026-05-12 Corroboration: cnbc.com
3. Google rolled "Gemini Intelligence" across every device you own
At the Android Show on May 13, Google packaged its 2026 AI features into a single brand it's calling Gemini Intelligence. The features ship across Android phones, Pixel devices, the Galaxy S26, Chromebooks, Google Home, GM cars from 2022 and newer, and the new Googlebooks laptop line. The most useful feature is Rambler, a voice-to-text upgrade in Gboard that strips the "ums," "ahs," and "likes" from what you dictate. Magic Cue and Now Nudge surface what you need before you ask, pulling context from your calendar, email, and recent searches. What this means for you: if you own a Pixel, Galaxy, or Android phone bought in the last 18 months, look for the Gemini Intelligence rollout in your settings this week. The feature most worth turning on first is Rambler. Voice-to-text that quietly removes filler words gets you to a clean email or note in roughly half the time.
Source: wired.com | Published: 2026-05-13 Corroboration: tech.yahoo.com
4. A new lab raised $650 million to build AI that rewrites itself
Recursive Superintelligence emerged from stealth on May 13 with $650 million in funding at a $4.65 billion valuation. The round was led by Google Ventures and Greycroft, with Nvidia and AMD also writing checks. The founder is Richard Socher, the AI researcher who built You.com and worked on the original ImageNet project. The pitch is the most ambitious in AI right now: build models that improve themselves without human retraining. The lab plans to ship its first product in 2027. What this means for you: the phrase "self-improving AI" sounds either thrilling or alarming depending on which podcast you listen to. The honest summary: nothing changes for your work this week. But the next time someone in a meeting says "AI capabilities are slowing down," you have a $650 million counter-example with Nvidia's name on it. Worth knowing.
Source: techcrunch.com | Published: 2026-05-14 Corroboration: siliconangle.com
5. OpenAI launched Daybreak to find security holes in your code before hackers do
OpenAI launched Daybreak on May 11, an enterprise cybersecurity platform built on GPT-5.5. It's the company's direct answer to Anthropic's Mythos, and it's available to companies, governments, and EU institutions in limited preview. The same week, Google reported it had thwarted what it called the first AI-driven mass exploitation event, where attackers used a public model to find and chain together zero-day flaws across hundreds of companies in hours instead of weeks. What this means for you: if you work at any company that runs its own software, or sells software, this is the year when the attackers got a free upgrade and the defenders got an expensive one. Ask your CTO this week which AI defense tool your company is buying. If the answer is "we're still evaluating," that's the answer of a target. If it's "we're piloting Daybreak or Mythos," that's the answer of someone who reads the news.
Source: macrumors.com | Published: 2026-05-11 Corroboration: cybernews.com
6. Anthropic and the Gates Foundation put $200 million toward AI for the rest of the world
Anthropic and the Bill and Melinda Gates Foundation announced a four-year, $200 million partnership on May 14. The money goes to AI tools for global health, education, and agriculture, with a focus on African languages, smallholder farmers, and underserved U.S. communities. The funding mix is grants, free Claude API credits, and engineer time. The same week, Anthropic disclosed its $44 billion annualized run rate, which made the announcement read less like charity and more like a signal that the company has decided what kind of business it wants to be. What this means for you: if your company has a corporate-social-responsibility line item that includes AI, this is the partnership template to copy. The interesting structural detail is the API credit component. Free compute is the new philanthropy currency. If you're a non-profit professional reading this, that's a credible path to ask any frontier lab for compute the same way you'd ask a tech company for a software license.
Source: anthropic.com | Published: 2026-05-14 Corroboration: forbes.com