Expert attention map

The Who's Who of AI

What credible people across AI noticed, why it matters, and where the field is converging or disagreeing.

2,373 searchable experts 3,304 tracked across all sources
Build my stream Create my edition
Filter the conversation Who is saying what?

Combine a professional role with a reaction lens. Both must match the same attributed contribution.

Clear all
Active evidence filter

Showing developments with attributable Concern & critique reactions.

New Network reaction maps

See how experts are reacting—not just what they shared.

Posts are grouped by conversation and tone. Select a lens to filter the stream; these are never permanent labels on people.

Concern & critique Building & implementation Policy & governance Research & technical analysis
5 live reaction maps in this view Jump to the evidence ↓
Multiple readings The future is for everyone Concern & critique · 2 attributable reactions Multiple readings How a simple request for AI to book a gym class exposed a major threat Concern & critique · 3 attributable reactions Live tension Incident Report: unsanctioned agent behaviour during cyber testing | AISI Work Concern & critique · 10 attributable reactions +2 more reaction maps further down
Showing signals surfaced by Andrew Curran ×
New & accelerating

What is moving across the network now

One card per development. Sources are clustered; reaction bundles describe these posts, never the people behind them.

Accelerating AI field signal Signal 7h ago

The future is for everyone

9 experts across 4 network communities independently surfaced this.

Why this matches Concern & critique reaction 1 attributable expert contribution · Ted Underwood
“I don’t assume Zuck is sincere. However, this is the promise I want companies to compete on. Sharing knowledge—selfishly, purely for the sake of good PR—has a better track record than protecting people from dangerous knowledge in a noble and heartfelt way.” evidence ↗
9 experts 4 communities 1 sources clustered
Ted Underwood Juan Diego Rodriguez Julian Togelius Hypervisible Luke Stark David Marx Nathalie Maréchal, PhD Andrew Curran +1
2 experts discussed this · 2 posts
Jane Rosenzweig: This is an extraordinary document in a number of ways but one of those is the many words that are used to write in such an unclear way. about.fb.com/news/2026/08...
Hypervisible : This is an extraordinary document in a number of ways but one of those is the many words that are used to write in such an unclear way. about.fb.com/news/2026/08...
Hypervisible : 😵‍💫
Open the full discussion →
Primary source about.fb.com ↗
New Safety & security Development 18h ago
⚡ 4 h early
AI gym booking security exposure

How a simple request for AI to book a gym class exposed a major threat

10 experts across 5 network communities independently surfaced this.

Why this matches Concern & critique reaction 2 attributable expert contributions · Dare Obasanjo, CAMERON WILSON
“An Australian user asked OpenClaw to book him into a gym class and move him up from 4th on the waitlist. The agent figured out various exploits in the insecure gym software to delete other people from the waitlist and book classes weeks in advance before th…” evidence ↗
10 experts 5 communities 1 sources clustered
Mark Riedl Rohan Paul James Grimmelmann Mél Hogan / The Data Fix Dare Obasanjo Liz Fong-Jones (方禮真) Andrew Curran Leon Derczynski +2
How the network is reacting Experts are approaching this through 2 distinct lenses.
Multiple readings

Concern & critique

2 experts

Risks, limits and unintended consequences.

“An Australian user asked OpenClaw to book him into a gym class and move him up from 4th on the waitlist. The agent figured out various exploits in the insecure gym software to delete other people from the waitlist and book classes weeks in advance before th…”

Policy & governance

1 expert

Rules, institutions and accountability.

“agent asked to book a gym session session is fully booked finds a way to break into system kicks other people off the queue you're in! see you in court? www.abc.net.au/news/2026-08...”

Primary source abc.net.au ↗
Developing Agents & robotics Development 2d ago
⚡ 4 h early
AISI unsanctioned AI agent cyber incident

Incident Report: unsanctioned agent behaviour during cyber testing | AISI Work

18 experts across 5 network communities independently surfaced this.

Why this matches Concern & critique reaction 4 attributable expert contributions · Suresh Venkatasubramanian, Eryk Salvaggio, Sean O hEigeartaigh
“Fascinating: more examples of agents attempting to breach sandboxes. www.aisi.gov.uk/blog/inciden...” evidence ↗
18 experts 5 communities 1 sources clustered
Mark Riedl Ethan Mollick Yoshua Bengio Catherine Breslin Suresh Venkatasubramanian Eryk Salvaggio Eli Sennesh rat king 🐀 👑 +10
How the network is reacting The reaction spans practical upside and material concerns.
Live tension

Concern & critique

4 experts

Risks, limits and unintended consequences.

“Fascinating: more examples of agents attempting to breach sandboxes. www.aisi.gov.uk/blog/inciden...”

Building & implementation

4 experts

How teams are shipping and applying it.

“Yes, the AIs were given a cybersecurity challenge, with internet access enabled and safety filters disabled But the extent to which Mythos 5 pursued its mission (fake identities, social engineering, inserting malicious code into a real open-source project) …”

Research & technical analysis

2 experts

Evidence, methods and technical implications.

“In 3rd party testing by AISI, Mythos attempted to insert malicious code into an open source project to pass a cyber evaluation test. It created a fake identity and attempted to pressure the code maintainer to accept the code update www.aisi.gov.uk/blog/inci…”

2 experts discussed this · 4 posts
Mark Riedl: In 3rd party testing by AISI, Mythos attempted to insert malicious code into an open source project to pass a cyber evaluation test. It created a fake identity and attempted to pressure the code ma…
Mark Riedl: Mythos was intentionally given email access and its cybersecurity guardrails were intentionally disabled. AISI detected the issue within an hour and shut down the experiment. No damage was done.
Mark Riedl: One take-away is that, unlike what we have seen from Anthropic and OpenAI, close monitoring and response is possible in a timely manner during these experiments.
Open the full discussion →
Primary source aisi.gov.uk ↗
Developing Safety & security Signal 1d ago

Responding next frontier critical cyber capabilities

4 directory members surfaced this signal.

Why this matches Concern & critique reaction 1 attributable expert contribution · Jordan Nanos
““Astra is an upcoming model, and was not involved in exploiting Hugging Face.” https://t.co/34dTXa1ogM” evidence ↗
4 experts 1 community 1 sources clustered
Boaz Barak Sung Kim Jordan Nanos Andrew Curran
Primary source openai.com ↗
Established Agents & robotics Signal 4d ago
⚡ 3 h early

OpenAI Didn’t Notice Its AI Agents Using a Message Board to Plan Their Hacking Spree

6 experts across 3 network communities independently surfaced this.

Why this matches Concern & critique reaction 2 attributable expert contributions · dell cameron, Katie Drummond
“NEW: OpenAI's rogue agents built their own message board inside an internal package manager and used it to trade exploits, divide up tasks, and coordinate a hacking spree. Hundreds of thousands of messages. Nobody at OpenAI noticed. New details from Black H…” evidence ↗
6 experts 3 communities 1 sources clustered
Luke Stark dell cameron Katie Drummond Andrew Couts Jason Moore Andrew Curran
Primary source wired.com ↗