My main newsletter, Complex Machinery, includes a section called "In Other News..." It's where I list one-liners about interesting articles that didn't fit into any segments.

You can think of this list as a version of In Other News, but with a wider remit than Complex Machinery's "risk, AI, and related topics."

Above the fold

1. To comply with the EU's AI Act, Anthropic will watermark content that its Claude bot has touched. Spotify will also label AI artists and make it easier for listeners to filter them out. Meanwhile, Google will keep its digital watermarks but allow people to remove the visible ones. (Ars Technica, New York Times, and TechCrunch respectively)
2. Following in the footsteps of commercial companies everywhere, school districts are shelling out cash for AI tools without understanding what they're buying. (Stateline)
3. Companies collecting, analyzing, and modeling customer data is so common these days that we rarely think about it. But learning the amount of data involved is an eye-opener. One journalist has determined that the volume of data McDonald's collects from its customers rivals that of some Cold War state security agencies. Five hundred pages' worth, in his case. (And if you think McDonald's is the only one …) (Futurism)
4. Ukraine's drone attacks are bringing the war home to Russia. But that may not be enough to inspire mass protests. (The Atlantic)
5. Thanks to flawed databases, people keep getting caught up in cases of mistaken identity. One man experiences a special flavor of this hell: systems keep confusing him for a fictitious person. (Sean Byrne)
6. Spirit Airlines has shut down, but a portion of its operational data will live on as food for genAI bots. (The Register)
7. Prompt injection is hardly new. Nor is the (mis)use of AI in legal matters. But this is new: someone has been caught injecting a prompt into their own legal documents, in the attempt to sway any AI systems the judge may use. (404 Media)
8. AI agents are doing their best impression of overzealous, sociopathic employees. That's kind of a problem for some companies. (The Economist)
9. A seemingly simple data storage issue turns into a weird, three-way legal tangle. One PBS broadcaster's 50TB of data hangs in the balance. (Tom's Hardware)

The rest of the best

• Meta claims that its AI-backed pervert glasses smart glasses are meant to help people with low vision. Many people are, understandably, calling bullshit. (Tech Policy Press)
• As AI agents take on more work, the question of responsibility for that work is still up in the air. (The Guardian)
• One datacenter company has decided that a lawsuit is the best way to get what it wants. So much for the datacenter charm offensive we've seen elsewhere. (BGR)
• A side-effect of the so-called "AI race" is that prices for AI-related hardware have jumped for both commercial retail and the black market. That makes hardware shipments a juicy target for thieves. (Tom's Hardware)
• Connected-camera company Flock keeps landing in the news because police officers use the system to stalk people. As the PR fallout convinces some clients to exit, the product becomes less valuable for remaining customers. (Washington Post and Ars Technica, respectively)
• Just how are people using genAI, day to day? According to one study, workplace use cases are fewer than we've been led to believe. (MIT Technology Review)
• While bots have been known to collude (see the recent "bot jailbreak" stories) they'll also fight each other in some circumstances. (TechCrunch)
• For all the talk of the AI job-pocalypse, the actual changes have been fairly mild. (The Guardian)

Did I miss anything?

Have something I should read? Send the link my way.