Sunday, October 25, 2026

Michelin denies using AI to write the food guide Texas paid millions for (This week in AI)

"Michelin declined requests to interview Guide leadership and declined an opportunity to review the evidence before publication."

[image or embed]

— Dave Infante (@dinfontay.com) October 19, 2026 at 7:17 AM
("When asked for comment, a spokesperson told D Magazine that the Michelin inspectors develop the descriptions based on their firsthand dining experiences and that the Guide has its own editorial team, but denied using AI to write or edit its restaurant descriptions.") ("The agreement that brought the Michelin Guide to Texas has expired after three years and $2.7 million in tourism funding, leaving the future of the prestigious restaurant ratings in Dallas and four other major Texas cities unsettled")





















gray lady covers gray paint www.nytimes.com/2026/10/22/t...

[image or embed]

— Ryan Mac 🙃 (@rmac.bsky.social) October 22, 2026 at 3:27 AM












Here's an update on my story in August about scientists using AI to make viruses. Now a startup has put AI viruses in a patient, to test them as a treatment called phage therapy. Gift link: nyti.ms/47pOLJv

[image or embed]

— Carl Zimmer (@carlzimmer.com) October 20, 2026 at 5:46 AM
(The report has not been peer reviewed or published in a scientific journal")











truly a wild time to be alive when the cryptography world is split between those worried about quantum computers and those whose private key is “4”

— Molly White (@molly.wiki) October 20, 2026 at 4:02 PM
("Security personnel around the world are on high alert Tuesday following the discovery of eight TLS certificates with private keys so weak they could allow anyone to mint trusted browser and app credentials for any site on the Internet") (“Explaining to my wife how there’s this one guy. And he alone is responsible for most of the world’s distribution of PHP,” a software developer ... wrote on X. “And if anything happens to this guy, that’s just it. Everyone is screwed.” He was referring to Surý.") (FWIW, "There isn't an easy exploit for this PHP vulnerability")











In September, the AI company Anthropic claimed its models were approaching “superhuman capabilities for geolocating outdoor photos”. Bellingcat put that claim to the test by benchmarking the performance of Anthropic and other LLMs. Here’s what we found: www.bellingcat.com/resources/20...

[image or embed]

— Bellingcat (@bellingcat.com) October 22, 2026 at 4:32 AM
(They've improved)











The first genuinely honest ad for generative AI I've ever seen, from Microsoft / CoPilot A mediocre white guy sees his non-white colleague getting praised for good work, he asks AI to *literally copy it*, and then he gets all the praise instead: www.ispot.tv/ad/SoDi/micr...

[image or embed]

— Ketan Joshi (@ketanjoshi.co) October 21, 2026 at 12:12 PM

Also hat tip to Jess Lawrence and a few others on Threads who saw the ad and posted about it www.threads.com/@jess_lawren...

[image or embed]

— Ketan Joshi (@ketanjoshi.co) October 21, 2026 at 12:37 PM

The amazing thing is that he isn't getting Copilot to do it from scratch - it is very much the direct plagiarism thing I have been talking about!!! bsky.app/profile/keta...

[image or embed]

— Ketan Joshi (@ketanjoshi.co) October 21, 2026 at 12:37 PM
(It's the small type at the bottom of the screen as he starts prompting)



















Publishers have been urged to delete Facebook, Whatsapp and Instagram accounts before a T&Cs change kicks in. 'Shameless land grab' would see Meta claim AI rights for everything publishers produce (whether it is on Meta platforms or not). pressgazette.co.uk/news/faceboo...

[image or embed]

— Press Gazette (@pressgazette.co.uk) October 19, 2026 at 3:34 AM












holy crap

[image or embed]

— Jessica McKenzie (@jessimckenzi.bsky.social) October 21, 2026 at 10:05 AM
(Didn't read it)











The entertainment giant hopes to get trust and safety tech, among other IP, from a startup whose chatbots dozens of states allege encouraged minors to self-harm and kill their parents.

[image or embed]

— WIRED (@wired.com) October 19, 2026 at 2:02 AM
("Disney Wants to Hire Most of the Employees From Bankrupt Character.AI")











Whatever the worst thing you can imagine being done with AI is, it turns out it's actually happening futurism.com/artificial-i...

[image or embed]

— jon christian (@jonchristian.net) October 21, 2026 at 6:31 AM
























Reagan appointee, born 1947











this dude already deleted his account lmao www.pcgamer.com/software/ai/...

[image or embed]

— Ty Schalter (@tyschalter.com) October 19, 2026 at 4:59 PM
















(Unclear if Anthropic has removed this feature from Claude)



































If you’re an author who’s been receiving verbose AI scam emails on a daily basis, and you’re wondering how gullible and ego-driven people think you are, this article helps to explain them. Sadly, they are successful enough to have been built into a big business. www.thedial.world/articles/new...

[image or embed]

— Oisín McGann (@oisinmcgann.bsky.social) October 19, 2026 at 11:51 PM




I had a renowned author contact me telling me how much he loved my book. After a few emails with him I realized I was talking to a AI that someone was training to write my own books.

[image or embed]

— Jack Murphy (@jackmurphyrgr.bsky.social) October 20, 2026 at 8:58 AM
























("Fighting game legend SonicFox pulls out of gaming event over AI-generated art")











AI bros out here burning tokens instead of just slapping an Arduino on a prayer wheel and making enough merit for the whole community at a fraction of the cost

[image or embed]

— Brendan O’Kane (@bokane.org) October 24, 2026 at 5:48 AM


i have 75 agents praying to send this guy to hell, effectively cancelling out his prayers

[image or embed]

— leon (@leyawn.bsky.social) October 24, 2026 at 6:43 AM