Issue 3 of the China AI Bulletin is out! This edition: China finalizes its Human-Like AI Interim Measures, DeepSeek and Moonshot launch new flagship models, and the NDRC blocks Meta's $2B Manus acquisition. Read it here: chinaaibulletin.substack.com/p/china-ai-b...

So excited to present a new resource for the #AIgovernance community, the China AI Bulletin: chinaaibulletin.substack.com

📬 Issue 2 of the China #AI Bulletin is out! This edition: Alibaba breaks with open-weight tradition with Qwen 3.6-Plus, Zhipu open-sources a model claiming #1 on SWE-Bench Pro, MIIT extends #AIethics regulation to the research stage, and TC260 launches a coordinated push on agent security standards.

📬 What if we Claudified the US constitution? I rewrote the US Constitution in the style of Claude's—turns out Madison and Anthropic have very different theories of control. 🧵

There is no faster route to self-knowledge than measuring the pixel distance between your pupils for a Chinese visa photo. Unfortunately, that knowledge is just a detailed inventory of all your minor facial asymmetries.

🔍 WR 75: A really bad week for xAI — Highlights from the latest Weekly Reckoning. Major setbacks for Elon Musk’s xAI/Grok and a broader shake-up in the AI world. 1/

The Weekly Reckoning is back! Get briefed on the tech news of the week (Grok's latest meltdown, China's new AI companion law, and insider trading in prediction markets), plus some musings on the excellent podcast BETH'S DEAD. 🔗👇

Slightly longer hiatus than planned 🫣 but the ER is back with a new edition on the Digital Vulnerabilities in the Age of AI (DIVAS) summit. Curious about how AI makes us vulnerable in new contexts? Check it out here: ethicalreckoner.subs...

My new indicator of AI-generated academic text is slightly random three-word groupings at the end of paragraphs. Ex: "XYZ is important for legibility, visibility, and control."

If things have been a little quiet around here recently, it's because I've been working towards this (slightly underwhelming) email... but now, it's #disserdone!

Dissertation update: word and zotero have teamed up to stage an eleventh-hour rebellion. They shall be brought to heel. #academia

We're getting there. #thesislife

Pleased that my and @floridi.bsky.social's article on US state #AI legislation is an Editor's Pick in the latest issue of the Journal of AI Law & Regulation! Want to understand the major tensions in AI regulation? 🔗👇, + our other piece on the US AI Action Plan. #STS #philtech @yaledec.bsky.social

🚨 new preprint from me and @floridi.bsky.social! We put the #AI Action Plan in context—how is it a continuation of existing US #AIgovernance efforts, what’s new, and what does it mean for China competition? 🔗👇 @ssrn.bsky.social @yaledec.bsky.social #academia #sts #philtech

🚨 Interested in infrastructure ethics? (Wait, don’t go!) New preprint “Extended Reality as a Fundamentally Mediated Reality” argues that XR isn’t just a cool tech, but infrastructures that shape our realities ➡️ urgent questions about equity, governance, and ethics. 🔗👇

📬 Anthropic settles its copyright case, Google avoids a break-up, and Grok swings right. Plus, why the EU needs a DeepSeek, not a Mistral. All this in today’s newsletter: ethicalreckoner.subs... #philtech #sts #newsletter

1. If passengers book a flight from the EU and it is canceled or significantly delayed, they are entitled to cash compensation — $300 to $700. In the US, flyers were supposed to get similar rights soon. But those plans have been scrapped by the Trump administration. Here's how it happened. 🧵

Anthropic is very, very relieved about this. Max penalties would have been extinction; this is <1% of their valuation.

#ICYMI, a post on #AI companies from last year that's just as relevant today with the continuation of the lawsuits, employee poaching, and sky-high valuations! #philtech #sts

This should have said "an inevitable future business model"--because we all know ads are coming 🫠

Recently have been seeing the "attention economy" narrative applied to #genAI--ie, that gen #AI companies are trying to maximize engagement to maximize profit. But I'm not convinced this is the case. #tech

I think my junk conference invites have peaked. #IEvenCompost

Word, you don't rest till I do. #DissertationLife

Maybe peer review takes so long because they're trying to fax invitations. #academia

#AIsafety collaboration could be a gateway to broader #AIgovernance cooperation between the US and China. Thinking a lot about cooperative levers these days... time.com/7308857/chi...#AIsafety#AIgovernance#AI

Some "creative" ingredients in this Google shakshuka rundown... I can't say I'll be adding "mix" or "your fingers" (not mine, yours).

#ICYMI: The week's tech news you need to know--including what's behind "AI psychosis," Meta's shocking guidelines allowing "sensual" chats with children, and how AI is helping design antibiotics. Plus, the opening bid for Chrome is in (no, it's not for sale... yet) 🔗👇

ICYMI: ChatGPT-5 murdered a bunch of AI "boyfriends" (and girlfriends). That and more in this week's newsletter: ethicalreckoner.subs...

I don't think my local coffee shop set out to be the perfect place to distill all my thoughts about the AI Action Plan, but they somehow have become that.

Two pieces for your Monday enjoyment! First, the weekly newsletter, featuring UK #OnlineSafetyAct age gates, Anthropic's anti-evil chatbot work, and a special guest contribution arguing that a recent court case might make NFTs relevant again 🔗👇

Now fascinated by #tsunami science. How can we have such granular understanding of hurricanes and earthquakes themselves but have such little idea of how many waves/how big they'll be? Is it a lack of observation infrastructure?

What happened in the #Anthropic and #Meta copyright cases? Is fair use the fairest way to decide if training LLMs on copyrighted material is ok? This week's newsletter digs into these questions 🔗👇 #techethics #philtech #AI

Between this very cool EMG wristband and the news that Ray-Ban Meta sales have tripled, it's a good day to be Meta XR.

Great translation from @jeffreyding.bsky.social of ChinAI confirming what we all kind of suspected: the DeepSeek integration wave was largely hype. chinai.substack.com/...

Four days in SF by the numbers: - Fun street sculptures: 5 - Hoverboards: 3 (apparently they’re still a thing?) - Ray-Ban Metas (noticed): 3 - Cybertrucks: 2 - “More espresso, less depresso” signs: 2 (get creative, guys!) - Men declaiming about TikTok witches and voice-to-skull communication: 1

The difference between "business casual" in the Bay Area and DC is staggering. Basically👖 vs 🎩

Another AI Action Plan contradiction "reject radical climate dogma"... but "embrace new energy generation sources" (geothermal, fission, fusion).

US in AI Action Plan: "Impose strong export controls on sensitive technologies." US in rare earth negotiations: "Sure, have some Nvidia H20s." Still reading through the full AI Action Plan, but... off to a great start?

The New England mind (mine) cannot comprehend summer weather that isn’t soup (SF)

📬 Today's WR: Meta's kicked off AI researcher draft season, China drops a second “DeepSeek moment,” hosting sites deepfake model whack-a-mole, and Android phones quietly become earthquake sentinels. Finally: why are people falling in love with AI chatbots? 🧵⬇️

maybe if Humane AI had put their product inside a Labubu instead of a pin people would've bought it

New Haven is trying to set the world record for largest (a)pizza party. What constitutes an official pizza party? Mandatory observed consumption of two slices (crusts excluded). What's the official definition of the crust-slice border? ¯\_(ツ)_/¯

🚨 How can #OpenSource principles keep pace with the rise of generative #AI? A new #preprint from the @yaledec.bsky.social proposes the Contextual Copyleft AI (CCAI) license — a new way to extend copyleft protections from open source code to the models trained on it. 🧵⬇️ #sts #philtech #OSAI

📬 This week's newsletter: Grok goes max hate speech (again), the EU's new AI Code of Practice, and whether #AI helps coders (hint: it depends). Plus, why I *want* my content to be scraped... but not everyone will. 🔗👇 #philtech #sts

I know I'm always on about how the publishing process takes too long, but I've never seen a more terrifying sight than this.

Labubus are the most brilliant application of hype culture and artificial scarcity that I’ve ever seen. These things are cost like $4 to make, are selling for $80-1800+, and you gotta collect 'em all.

Was a delight to chat with Akky! 🔗👇

🚨 How can we overcome the implementation gap and realize #AI’s potential benefits in global health? Spoiler alert: it’s not just more technology. We point out five fundamental infrastructure issues that need to be overcome for ethical AI implementation in global health. #techethics #academia #paper

Paper submitted... and it only took four hours because I had to figure out the silent ORCID linkeage failure--turns out it was linked to another account in that system. Which journal and which email? We're not gonna tell you 🙃 #academia

The state AI law ban is dead 🪦 Bad news: we have to revise our paper on state AI legislation Good news: states will be able to legislate key state-specific AI issues Bad news #2: we still won't have comprehensive federal AI legislation