AI Reading for Saturday June 13

Anthropic blocked all customer access to Fable 5 and Mythos 5 after the Commerce Department ordered them to suspend access for all foreign nationals globally, including foreign national Anthropic employees; the duration of the restrictions was not specified. - The New York Times

Did Pliny the Liberator's jailbreak, combining decomposition to hide intent, plus character-substitution homoglyph attack, lead to Anthropic order? Other sources reference Amazon research, which sounds different. - VentureBeat

Critics called the ban "cartoonish" and drew parallels to 1990s encryption export battles; UK officials expressed concern over technological sovereignty - Business Insider

UK politicians, including AI minister Kanishka Narayan, reacted with concern about technological sovereignty, pointing to the government's £1.1bn AI chip investment; critics argued the UK has "virtually zero sovereign AI capability." - City AM

Tech blogger argues the Fable ban reflects Trump administration political targeting of Anthropic rather than legitimate national security regulation. - 12gramsofcarbon.com

I thought Dario Amodei's latest was good and thoughtful and nuanced but also had more hope (naiveté?) than I can muster for positive democratic and international regulation of AI, that plus the breathless hype about Fable/Mythos put a target on his back from the "f*** democracy" faction. we just want oligarch race war ragebait and populist vibes that we can control, not stuff well-grounded in facts and reason.

"It's literally the gulag." A Meta engineer describes their job in the new 6,500-person "Applied AI" unit, where highly-paid Silicon Valley talent are now "draftees" forced to write puzzle problems to feed the beast. - WIRED

Someone interrupted a livestreamed, employee-only presentation at Meta earlier this week with an expletive-filled outburst about “being the company’s bitch,” according to a recording heard by WIRED. The individual then asked the people leading the call to write to a specific Meta AI executive and "tell him that he's a piece of shit."

Zuckerberg acknowledged Meta made mistakes in its AI workforce restructuring and pledged more stability as the company avoids further layoffs this year. - Reuters

Kill switch where? OCC and Federal Reserve are questioning banks about AI governance, vendor risk, and kill switches in routine examinations as AI adoption accelerates. - Yahoo

Multi-state attorney general coalition, led by New York, subpoenaed OpenAI for records on data handling, minors, and chatbot sycophancy, fueling concern about OpenAI's mounting legal exposure as it prepares its IPO - Engadget

Illegal AI Overviews: Munich court rules Google liable for false statements made by AI Overviews, setting potential global precedent. - WIRED

Study finds GPT-5.2, Gemini 3.1 Pro, and Claude Opus 4.6 outperform specialized clinical AI tools OpenEvidence and UpToDate on medical benchmarks. - Nature

Reader letters about asking ChatGPT medical questions. - The New York Times

a bit concerned about the forensic (legal?) psychologist who uses it to ‘arrange’ findings. at a minimum that seems likely to be brought up when challenging those findings.