James Titcomb retweeted
I understand Andy Burnham is taking the unusual step of bringing his AI minister, Kanishka Narayan, to this year’s UN summit in New York tomorrow.
At the same summit, Sam Altman is briefing the UN Security Council. Expect a week of discussions about risks vs rewards from AI models.
1
11
131
14,372
Why do AI people love this comparison so much? Millions of people die from food poisoning
Artificial intelligence is "less regulated than selling a sandwich" in the UK and US, an AI safety campaigner tells @CathyNewman.
The King will host AI leaders in Scotland this week amid dire warnings about the risks of rapid AI development.
📺 Sky 501 and YouTube
11
5
21
9,939
This has been Anthropic's holding line on non-US Mythos access for months now and it doesn't seem like much is happening
1
299
'Thank you for coming to me'
Exciting day for NVIDIA and @huggingface.
Open models strengthen safety and cybersecurity, accelerate innovation and diffusion, and enable sovereignty. They allow every developer, startup, university, industry and country to build with, customize and benefit from AI.
Thank you @ClementDelangue for coming to me.
NVIDIA is going to be a great home for Hugging Face, its community and the future of open models. 🤗
blogs.nvidia.com/blog/nvidia…
1
332
Hard to see any problems with the "AI will defeat bioterrorism by developing new vaccines that everyone will take" argument
3
305
AI Security Institute responds to Frontier Security's claim that Chinese model Kimi exploited a loophole in AISI tests.
"These claims are inaccurate and irresponsible"
1
2
335
James Titcomb retweeted
NEW: OpenAI gives first detailed debrief of the Hugging Face incident at Black Hat conference
In a session I attended today at Black Hat, OpenAI's Eric Wallace and Michael Dalton said the company is "consciously slowing down research to enhance security" while a full technical postmortem is still underway.
* OpenAI traced the roots of the attack back to May 7, during training of an unreleased frontier model—not July.
* The most surprising detail: AI agents accidentally created an internal message board, allowing separate evaluation runs to share exploits, discoveries and work assignments.
* OpenAI said it shut the message board down after an internal security incident—only for the agents to independently recreate it days later using a different communication method.
* OpenAI called the incident a "watershed moment" for AI security and warned that "agent orchestrated fully automated offensive attacks are real now."
* The company also said it is "consciously slowing down research to enhance security" while overhauling its defenses.
groundlevel-ai.com/p/openai-…
127
331
1,834
2,252,473
AI safety vibe shift has been so fast. AISI only ran this test a couple of weeks ago but this setup looks pretty loose now.
“we did not revisit this judgment quickly enough as capabilities advanced”
1
1
313
Feels like a lot of the problem here is Apple's policy of requiring employees to use their personal iCloud account for work openai.com/index/apple-is-ge…
3
775
$2,000 of inference to solve decades-old maths problems
but $1,000,000s of inference from people asking ChatGPT to explain what High-dimensional sphere packing is
ten significant advances in mathematics and theoretical computer science.
solved using an internal version of Astra, our next major model, for a total cost of about $2000 at Sol API prices:
1
1
778
Google always pushes back on the idea that search has got worse over, but the search by date filter has been broken globally for the last few days and the company doesn't seem to have acknowledged it
1
378
Only one man for the job
Netflix Sued For Losing 'Master Copy' of Unreleased Nicolas Cage Movie ift.tt/fq3wrHa
336