Vert|Rule retweeted
The problem with credit chasing around LLMs is that essentially all of the components existed: transformers, next-word prediction, transfer learning, scaling trends. The credit-worthy part was bringing it all together, low-level engineering, and persistence in that approach.
Vert|Rule retweeted
I spent the last 2 days trying to figure out how scary Claude Mythos is.
I think it's fairly scary, though not because of the hacking:
1. It indicates fully-automated AI R&D is coming sooner
2. Its alignment seems better, which is good. But all the alignment tests have serious flaws, which is bad.
3. There are a few specific warning signs Mythos might not be trustworthy
I explain what stood out to me in the 244-page System Card and 59-page Alignment Risk Report in this essay for the 80,000 Hours Podcast.
Judging by those 2 reports Anthropic itself seems kinda scared of Claude now.
And I'm sure views vary widely within the company, but at times it feels like they only give themselves a 50/50 chance of being able to keep the next few Claudes fully under control.
So I guess we're on the same page!
If it does turn out we can take the safety results at face value we may look back and see this week as watershed good news. If they can't, the opposite.
(I wish I had had more time to look into how reassuring it is that their automated monitoring systems don't seem to be picking up much misbehaviour post internal-deployment. I think that's what someone at Anthropic who feels more relaxed would point to. Next time!)
Links below - enjoy!
nitter.gq/sleepinyourhat/status/…
Vert|Rule retweeted
“STEALING ISN'T INNOVATION” More than 700 leading figures from across the creative industries including Cyndi Lauper, Questlove, Chaka Khan and Scarlett Johansson back the @human_artistry campaign to prevent unauthorized use of AI stealingisntinnovation.com/
Vert|Rule retweeted
More than 700 actors, writers, and other creators signed a new anti-AI statement saying tech firms are building AI by taking copyrighted work without permission or pay.
It asks AI developers to use licensing deals and partnerships instead.
Scarlett Johansson, Cate Blanchett, and Joseph Gordon-Levitt are among the public names backing the campaign.
The campaign’s proposed change is a licensed input stream, meaning datasets come from contracts that specify what can be used, for what purpose, and under what payment terms, so training is tied to auditable rights.
---
variety .com/2026/film/global/scarlett-johansson-cate-blanchett-anti-ai-campaign-1236637571/
Vert|Rule retweeted
Earlier this year, after receiving emails from people in the throes of AI psychosis, I began to keep in touch with one man as he journeyed to recovery.
I'm grateful to James that he was willing to go on record. He wanted people to know: humanlike chatbots are dangerous.
Vert|Rule retweeted
At the TIME100 AI celebration last night, @nlyonne stood in front of the emperors of AI and challenged them to their face. Doing that takes guts. And I’m incredibly honored she cited my work as well.
Thank you for speaking truth to power, Natasha. 🙏
m.youtube.com/watch?v=RUu9Ov…
