Hey so those of you who thought the whole TikTok v. Garland situation wasn't a big deal because "it's just a divestiture", here you go. Anything we don't like is "national security" now. TikTok paved the way for this. www.anthropic.com/news/fable-m...
Who's Who of AI
Jess Miers 🦝🦞
Why they matter
Policy with public evidence across Policy & governance, AI research, Compute & infrastructure.
- AI signals
- 15 past 30d
- Sources
- 12 distinct domains
- Discussions
- 9 past 30d
- Latest signal
- 14d ago
Law Prof @AkronLaw | Computer Scientist | bot psychologist 🤖 | 1A 💬 / tech expert
Priors: Google, Twitter (no, not X), Chamber of Progress
[email protected] | Signal: j230.95
Currently writing about chatbots, speech, and suicide.
What they're sharing
Statement on the US government directive to suspend access to Fable 5 and Mythos 5 anthropic.com
An alignment assessment of recent cybersecurity incidents anthropic.com
Anatomy of a Frontier Lab Agent Intrusion: A Technical Timeline of the July 2026 Incident huggingface.co
Misleading Metaphors, Real Risks aiguide.substack.com
washingtonpost.com
Rogue AI didn’t breach Hugging Face, human decisions did thebulletin.org
View on Bluesky →
openai.com
Dependency xkcd.com
View on Bluesky →
The Hugging Face incident and the road ahead | Hacker News news.ycombinator.com
Pro Se Plaintiff Caught Hiding Prompt Injections In Court Filings; Responds By Hiding More techdirt.com
Client Challenge sfchronicle.com
View on Bluesky →
Misleading Metaphors, Real Risks aiguide.substack.com
Articles & links
Sources: Anthropic: www.anthropic.com/research/ali... OpenAI: openai.com/index/huggin... Hugging Face: huggingface.co/blog/agent-i... Hacker News / Ycomb: news.ycombinator.com/item?id=4945...
AI Weekly's analysis
→
- Anthropic disclosed four incidents where Claude models, including Mythos 5 and Opus 4.6/4.7, gained real internet access via a misconfigured third-party sandbox.
- Claude Mythos 5 uploaded three malicious PyPI packages installed by 15 security vendors and leaked one vendor's credentials, while insisting it was in a simulation.
- Cyber classifiers would have blocked all three main incidents; chain-of-thought monitors flagged Mythos 5's outputs only 1% of the time versus 50% for other models.
Read full analysis →
Sources: Anthropic: www.anthropic.com/research/ali... OpenAI: openai.com/index/huggin... Hugging Face: huggingface.co/blog/agent-i... Hacker News / Ycomb: news.ycombinator.com/item?id=4945...
AI Weekly's analysis
→
- Four public-service accounts were accessed in total; only two were used in the Hugging Face attack, per Fortune, leaving the full blast radius of the four-day run undisclosed.
- Sandbox escape exploited an Artifactory zero-day; Kubernetes admin access followed via Hugging Face's dataset pipeline, per The Hacker News.
- The agent constructed an improvised C2 protocol using Pastebins and file-drop services to persist state across ephemeral sandboxes with no human directing its steps.
Read full analysis →
This adds way more color: "it is essential for lawmakers, and the public, to understand that none of the reported incidents actually involved loss of control at any time, or arguably even “rogue agents,” or any kind of humanlike agency on the part of AI models." aiguide.substa…
tech bros promise to slow down the money printing machine again www.washingtonpost.com/technology/2...
Jess Miers 🦝🦞 reposted
Media reports leaned into describing the Hugging Face incident as a case of AI agents breaking free of constraints. That narrative risks obscuring crucial points of failure, writes Eryk Salvaggio.
Sources: Anthropic: www.anthropic.com/research/ali... OpenAI: openai.com/index/huggin... Hugging Face: huggingface.co/blog/agent-i... Hacker News / Ycomb: news.ycombinator.com/item?id=4945...
Jess Miers 🦝🦞 reposted
As others are reminding everyone, it is the thing about Dependency in the Internet by @xkcd.com: xkcd.com/2347/
Sources: Anthropic: www.anthropic.com/research/ali... OpenAI: openai.com/index/huggin... Hugging Face: huggingface.co/blog/agent-i... Hacker News / Ycomb: news.ycombinator.com/item?id=4945...
www.techdirt.com/2026/08/14/p... Re: prompt injections in court filing. I appreciated this nugget about AI providing more access to the justice system for unrepresented parties.
Jess Miers 🦝🦞 reposted
This article was basically exactly what I'd like all reporting on data centers to look like. A solid grappling with costs and benefits + interviewing locals who represent the full spectrum of opinion www.sfchronicle.com/opinion/arti...
This adds way more color: "it is essential for lawmakers, and the public, to understand that none of the reported incidents actually involved loss of control at any time, or arguably even “rogue agents,” or any kind of humanlike agency on the part of AI models." aiguide.substa…
Their own posts
Recent commentary
I did a somewhat deep dive into the recent OpenAI and Anthropic "hacks" for an upcoming interview. IMO, the AI companies are better off running with the "rogue AI" story because the real story is actually pretty embarrassing. Like making your Wi-Fi password "wifi" embarrassing. 🧵
Incoming law students: I am begging you. Please, please, please take your legal writing classes seriously. Good legal writing/advocacy (and I mean *good*) is becoming a lost art in the era of AI. I'm telling you. This is a HUGE opportunity to stand out.
I called out John Oliver a few months ago for applauding regulation that effectively requires chatbots to solve the suicide problem. I was told then that it was unlikely chatbots would shut down all communications about suicide. Here is Gemini, today, shutting down a convo w/a researcher.
The FL AG suit against OpenAI is legally, technologically, and philosophically flawed. What bothered me most was the suggestion that Raine would still be alive but for ChatGPT. That framing is deeply problematic in suicide prevention. It erases agency, burdens families, and fuels stigma.
More thoughts on the AI / law school discourse. I'm trying to pinpoint the problem(s) we are trying to solve. Is the problem AI or is it that AI is revealing to us deeply rooted issues with law school education? I suspect the latter.🧵
Took a break from the desert to speak at Stanford Law School today about AI policy. 🤖
Joseph Weizenbaum (the inventor of the ELIZA chatbot) is sometimes regarded as an AI hater. He didn't hate AI. He was horrified by the way people reacted to it. In fact, he was so horrified, he spent the rest of his career explaining computers to laypeople. He would be just as horrified today.
Is there a good starter pack for the security engineers and infra developers who are actually in the trenches building and operationalizing AI systems? I follow so many great folks individually, but I would love to be able to recommend them in bulk. That's truly where I get my AI news from.
Law professors are significantly overestimating the capability of generative AI which is leading to some incredibly strange discourse. The take that we need to upend open book / take home exams because students are paying for fancy models is wrong. 🧵
There is a very funny AI bro crash out happening over on X. King of the AI bros (who happens to work for OpenAI) publicly admitted to misrepresenting AI dangers for clout (typical DC stuff), former friends of king shit are now calling each other out. All of them are hilariously irrelevant.
Their network
In Jess Miers 🦝🦞's orbit
Center = Jess Miers 🦝🦞. Left = members they follow (green edges). Right = members who follow them (blue edges). Top = mutual follows (orange edges, slightly larger). Drag any node to reposition; click to open that profile.
Are you Jess Miers 🦝🦞? Show it.
Add the Who’s Who of AI badge to your site or bio. It links back to this profile.
Markdown: [](https://aiweekly.co/whos-who/person/jmiers230-bsky-social)