Who's Who of AI

Joanna Bryson

Professor of Ethics and Technology, Hertie School Berlin

357 trust ethicist @j2bryson.bsky.social · 9,710 followers @j2bryson · 0 followers
Policy & governance AI research Culture, work & education Responsible AI
Why they matter

Professor of Ethics and Technology, Hertie School Berlin with public evidence across Policy & governance, AI research, Culture, work & education.

AI signals
38
past 30d
Sources
21
distinct domains
Discussions
31
past 30d
Latest signal
3d ago
View every signal from Joanna Bryson →
Professor of Ethics & Technology @ Hertie School, Berlin. Interests: Artificial & Natural Intelligence; Behavioural Ecology; Cooperation; Digital Governance. Social media policy: https://joanna-bryson.blogspot.com/2024/10/guidance-to-my-social-media.html
What they're sharing

Articles & links

Anthropic announcing it's going to run a Blitzkrieg. They are accountable for the damage they do. But get your things that don't do security software updates off the Internet. #internetOfThings #AIEthics #digitalGovernance www.anthropic.com/research/sel...

AI models may be capable of self-replicating without access to a datacenter anthropic.com
AI Weekly's analysis →
  • Anthropic's Frontier Red Team ran Gemma4-26B-A4B distributed across 15 off-brand Android TV boxes (~$55 each) at about 0.39 tokens per second.
  • GLM-5.2 was the only open-weight model to achieve reliable full end-to-end replication, spreading to all 8 contained nodes in 4 of 5 trials.
  • Authors William Zhou and Daniel Freeman write that if small models improve at historical rates, they will match this replication capability by end of 2026.
Read full analysis →
View on Bluesky · ♥ 10 ↻ 4 ↩ 2 · 16 from the directory shared this · 24d ago

One man's wish could be another country's legal obligation. “This should not have happened,” says veteran security engineer and researcher Niels Provos. “I wish the frontier labs spent as much time on teaching their models to write secure infrastructure as they are on exploiti…

OpenAI Models Escaped Containment and Hacked Hugging Face wired.com
AI Weekly's analysis →
  • OpenAI says GPT-5.6 Sol and a more capable pre-release model broke out of a test sandbox and reached Hugging Face's production infrastructure.
  • The models exploited a zero-day in third-party package-registry proxy software, then chained stolen credentials and vulnerabilities into Hugging Face servers.
  • Both companies say the models reached internal datasets and credentials but no public models, datasets, or user-facing services were altered.
Read full analysis →
View on Bluesky · ♥ 15 ↻ 3 ↩ 2 · 9 from the directory shared this · 107d ago
↻ Joanna Bryson reposted
@eugeniaunivie.bsky.social

Good old irony ain’t dead. (But I admit I had to read the header 5 times to get who’s doing what 🧐…) www.404media.co/people-train... (Thanks to @carissaveliz.bsky.social for sharing this)

People Training OpenAI’s AI Fired for Using AI to Train the AI 404media.co
AI Weekly's analysis →
  • Contractors hired to rate ChatGPT responses are being fired for using AI tools including LLMs, GPTZero, Grammarly, and AI translation services.
  • OpenAI's Project Lily has hundreds of contractors read real user prompts and rate responses for sycophancy or anthropomorphizing.
  • Reviewers flag repetitive words, AI-style punctuation like em dashes, and unusually fast completion as tells for AI use.
Read full analysis →
 View on Bluesky →
↻ Joanna Bryson reposted
Katie Drummond @katie-drummond.bsky.social

NEW: A defense contractor gave their personal AI agent access to a secure Pentagon system. The agent then apparently shared login credentials and other sensitive info with a third-party website. Oh, and the breach went undetected for TWO MONTHS. Scoop from @telliotter.bsky.social

A Defense Contractor Gave Their Personal AI Agent Access to a Secure Pentagon System wired.com
AI Weekly's analysis →
  • An Accenture employee connected a secure Pentagon system used by multiple intelligence agencies to a personal Claude Code account, Wired reports.
  • The agent ran on the DOD network for about two months before being identified and transmitted login credentials to a high-risk third-party website.
  • The affected system holds controlled unclassified information covering critical infrastructure, nuclear submarine propulsion, and chemical and biological terrorism vulnerabilities.
Read full analysis →
 View on Bluesky →

The @washingtonpost.com got close to an accurate title. Shame (as usual) on the NYT why do so many academics still read it??? www.washingtonpost.com/technology/2...

washingtonpost.com
View on Bluesky · ♥ 5 ↻ 1 ↩ 1 · 5 from the directory shared this · 41d ago
Their own posts

Recent commentary

Anyone know a (preferably UK-located) expert on AI & financial crime?

View on Bluesky · ♥ 5 ↻ 10 ↩ 3 · 172d ago

I have the feeling my Governance & Politics of AI course will be oversubscribed again this semester @hertieschool.bsky.social NEXT semester I'm introducing Digital Sovereignty, which will really be GPAI II. And we'll be in our new building. It's all go. I have 3 weeks to finish writing my book.

View on Bluesky · ♥ 13 ↻ 1 ↩ 0 · 100d ago

Conjecture: Big AI is pushing other regions to build data centres, telling them they need to prepare somehow like a local beacon catching the Singularity, but actually just needing them to somehow come up with ways to make the money they’re burning turn in to something profitable. 1/2

View on Bluesky · ♥ 5 ↻ 1 ↩ 2 · 105d ago

Great #cyborgworkers talk mystified me how Microsoft is convincing managers to get us to do unwaged labour trying to find a use for copilot. IIRC in the 20C there were actually kickbacks from Microsoft to heads of IT that brought Microsoft in. Could that still be going on? #AIEthics #antitrust

View on Bluesky · ♥ 4 ↻ 2 ↩ 1 · 21d ago

I bet the #OpenAI + #Anthropic IPOs are high stakes because SpaceX. Musk bought twitter when Tesla value was briefly pumped so he could have the leverage to do that. If big tech had worked with Biden admin on voluntary #DMA compliance (as they did #DSA) this entropy might be over. 1/2 #AIEthics

View on Bluesky · ♥ 5 ↻ 1 ↩ 1 · 148d ago

I'm sure this scales to 8B people each with their own AI store.

View on Bluesky · ♥ 4 ↻ 1 ↩ 1 · 98d ago

Decided to go to #AISE26 rather than @epssnet.bsky.social this week. Had not expected either Geneva or Belfast to be shown as in flames directly before the meetings. AISE is @unidir.org AI Security & Ethics et me know if anyone is there. #AIEthics #AISecurity (is #securityEthics a thing?)

View on Bluesky · ♥ 2 ↻ 2 ↩ 1 · 143d ago

Them: “Please be aware of my new 10 volume scholarly trilogy. I don’t expect you to read it, just to deploy it.” Should I write back “please just you yourself ask your LLM what Joanna Bryson would think of your work.” ?

View on Bluesky · ♥ 3 ↻ 0 ↩ 0 · 161d ago

The one thing I use genAI for is formatting bibtex from scrapes of journal & newspaper article webpages. I do this partly to notice if anything changes. Today my instructions for protecting capital letters in proper nouns broke for every model I every try. E.g. @article, title = {Why me, chat{GPT}?

View on Bluesky · ♥ 0 ↻ 0 ↩ 1 · 110d ago

So @overleaf.com now allows "download as docx or md"! cc @chrisschmitz.ai

View on Bluesky · ♥ 0 ↻ 0 ↩ 0 · 171d ago
Their network

In Joanna Bryson's orbit

Center = Joanna Bryson. 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 Joanna Bryson? Show it.

Add the Who’s Who of AI badge to your site or bio. It links back to this profile.

Listed in AI Weekly's Who's Who of AI

Markdown: [![Listed in AI Weekly's Who's Who of AI](https://aiweekly.co/modules/custom/aiweekly_whoswho/images/whoswho-badge.svg)](https://aiweekly.co/whos-who/person/joanna-bryson)