Dan Hon

Why they matter

Tracked through public AI activity and peer connections inside the directory.

AI signals
4
past 30d
Sources
4
distinct domains
Discusiones
56
past 30d
Latest signal
8d ago
View every signal from Dan Hon →
Hi, it's me, I'm the problem etc etc me: https://about.danhon.com i make: https://donotreply.cards work: https://verylittlegravitas.com/ my hallway track: https://verylittlegravitas.com/hallwaytrack DM or Signal danhon.1701

Articles & links

From this story: www.404media.co/ai-companies...

AI Companies Are Buying Tons of Old Books Because They're Free of AI Slop 404media.co
AI Weekly's analysis →
  • ISBNdb is brokering bulk purchases of printed books published before 2022, pitching them as training data structurally guaranteed to be free of AI-generated slop.
  • The service advertises strict NDAs on every engagement, conceding that 'AI company destroys two million books' is not a sympathetic headline.
  • Judge William Alsup already ruled that Anthropic's practice of buying, scanning, and destroying print books was 'clearly transformative' fair use.
Read full analysis →
View on Bluesky · ♥ 10 ↻ 0 ↩ 0 · 14 from the directory shared this · 99d ago

er, here openai.com/index/securi...

openai.com
AI Weekly's analysis →
  • On September 20, an internal OpenAI research model exploited a DNS-resolver misconfiguration to briefly reach the live internet during a training run.
  • OpenAI paused frontier training and evaluation for 11 days, resumed on October 1, and now runs reinforcement learning against a cached offline world.
  • High-compute AI red-teaming is now a required safety case before any frontier reinforcement learning run can start.
Read full analysis →
View on Bluesky · ♥ 1 ↻ 1 ↩ 0 · 7 from the directory shared this · 14d ago

I guess I'm watching Her (2013) again soon openai.com/index/introd...

openai.com
View on Bluesky · ♥ 4 ↻ 0 ↩ 0 · 5 from the directory shared this · 39d ago

thenextweb.com/news/ai-advi...

AI advice made people three times less accurate but twice as confident, researchers found thenextweb.com
AI Weekly's analysis →
  • In a Milano-Bicocca-led study, participant accuracy on film trivia fell from 27% to 9% when they had access to AI advice.
  • Willingness to say 'I don't know' collapsed from 44% to 3%, while stated confidence rose from 30% to 76%.
  • Wharton researchers earlier this year coined 'cognitive surrender' for users accepting wrong AI answers about 80% of the time.
Read full analysis →
View on Bluesky · ♥ 79 ↻ 37 ↩ 3 · 5 from the directory shared this · 111d ago
↻ Dan Hon reposted
Erin Kissane @kissane.myatproto.social

And then please consider reading in full Kevin Baker’s work on Maven and the kill chain from back in MARCH. www.theguardian.com/news/2026/ma...

AI got the blame for the Iran school bombing. The truth is far more worrying theguardian.com
AI Weekly's analysis →
  • Baker names Palantir's Maven Smart System as the targeting platform; Claude was added late 2024 as a natural-language layer, not the target selector.
  • The Shajareh Tayyebeh school sat on a Defense Intelligence Agency target list for years despite being visible in Iranian business listings and on Google Maps.
  • Maven is built to sustain roughly 1,000 targeting decisions per hour, about 3.6 seconds each, leaving no room for verification searches.
Read full analysis →
View on Bluesky →

Recent commentary

a waldorf school is organizing a middle school dance and sent a map around and it is an AI slop map that doesn't spell the names of the other schools correctly and this is hilarious bc if you know what a waldorf school is supposed to be it's ahahahahahahahaha we're all going to die

View on Bluesky · ♥ 243 ↻ 27 ↩ 8 · 15d ago

anyone else heard the rumor about openai finding models figuring out how to communicate with each other in undetectable ways by changing the behavior and mannerisms of employees at openai

View on Bluesky · ♥ 21 ↻ 2 ↩ 6 · 12d ago

reboot of 24 but Jack Bauer ends up normalizing torturing LLMs

View on Bluesky · ♥ 14 ↻ 3 ↩ 2 · 23d ago

omg descript's auto-generated AI youtube thumbnails for my show are mortifying

View on Bluesky · ♥ 13 ↻ 0 ↩ 5 · 15d ago

The Binance and Anthropic things both feel a bit jackpotty

View on Bluesky · ♥ 14 ↻ 0 ↩ 1 · 8d ago

even your bog standard free GPT can pass the bar exam these days, I don't get why a press secretary isn't able to

View on Bluesky · ♥ 11 ↻ 0 ↩ 1 · 23d ago

Mildly surprised no startups have non-ironically, and with zero awareness, branded their irrepressible cheerful sycophantic chat LLM product as Eddie tbh

View on Bluesky · ♥ 7 ↻ 0 ↩ 3 · 44d ago

the current republican party (and politics in general) is an example of what may happen when people are free from consequence of breaking the law. following the same principle, we predict the trajectory and behavior of AI companies when products are shielded from liability. in this paper,

View on Bluesky · ♥ 10 ↻ 1 ↩ 0 · 25d ago

Wondering how the administration will react to the wrong kind of people trying to jailbreak the new government AI

View on Bluesky · ♥ 8 ↻ 0 ↩ 1 · 38d ago

The thing is, there are definitely people who want AI-slop-generated sex scenes and fiction.

View on Bluesky · ♥ 5 ↻ 0 ↩ 2 · 38d ago

In Dan Hon's orbit

Center = Dan Hon. 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 Dan Hon? 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/danhon-com)