If only it were just a tool – but instead Claude 4 sometimes shows a stubborn mind of its own, complete with blackmail and snitching. Simon Willison read through the official documentation.

ChatGPT stifles idea diversity in brainstorming, a recent study says. Heads up though: researchers were looking at ChatGPT 3.5, which dropped in November 2022 and has since been discontinued.

The Pulitzer Prize is once again highlighting investigations that use AI – but we’re talking “regular” AI here, not ChatGPT. NiemanLab has the details.

Everything’s happening at once: Google’s showing off the future of search at I/O, Karen Hao’s book “Empire of AI” is out, and Anthropic is releasing Claude 4.

Google’s new AI chatbot mimics Perplexity and OpenAI’s ChatGPT, prompting questions about the future of information discovery. (Tripp Mickle, New York Times)

Create soulless and interchangeable content at the push of a button? There Is An AI For That: TrendlyAI.

Dan Shipper of Every startup uses AI to augment, not replace, writers – a contrarian take amidst AI-fueled media disruption. (Benjamin Mullin, New York Times)

Oops: MIT has withdrawn a paper that claimed productivity increases through AI. (Anthony Ha, TechCrunch)

It’s not about chatbots. What Big Tech is really working on: “embedding AI deeply into the fabric of their ecosystems, creating an invisible layer that spans our digital lives.” (Ezra Eeman made a nice visual, LinkedIn)

“The best way to get your client’s message into the output of ChatGPT, Claude, Gemini, and the rest is by talking to journalists.” (Ben Smith, Semafor)

Florent Daudens has built an open source version of NotebookLM’s podcast to break down the top AI research paper each day: “Fully automated, now live on Spotify.” (LinkedIn)

What does it actually mean to get your news from AI? Laura Preston laments endless, evasive bullet points. I don’t know what’s worse: journalists thinking they can ignore AI after this, or people reading AI content and just going “good enough.” (Columbia Journalism Review)

Podcast: Using ChatGPT, especially for therapy, may prove difficult to disengage from, as chatbots employ “personalization loops.” (Ryan Broderick, Panic World)

The Financial Times has a brutal takedown of Sam Altman and OpenAI based on a 22-second video of him cooking pasta.

AI search is (likely) hurting publishers: “The data is really messy, and … we have no real way of differentiating … to see what is really going on. But overall, it all points to there being a problem.” (Thomas Baekdal)