If you, like me, feel sick when you process what's actually happening here, @NoTechApartheid are leading organizing in the long tradition of scientists and tech workers standing up against violence machines.
SAFETY
-
Cloud AI Companies Implicated in Surveillance System Facilitating Mass Murder
By
–
There is almost no way the large cloud companies providing massive compute services AI surveillance & data processing APIs aren't implicated in the system of 'ad targeting for death' being used to facilitate mass murder in Gaza. This is a nightmare. And it's going to take
-
LLMs Vulnerable to Short Token Sequences Manipulation Study
By
–
This is an important paper that demonstrates empirically something that folks like @goodside & I have been warning about. LLMs are easy to fool, and this paper shows that short token sequences are enough to alter the advice of AIs about what product to buy. SEO for AI is here
-
Stanford AI+Policy Symposium Convenes Global Safety Institute Leaders
By
–
Next week, @StanfordHAI & @StanfordCyber will convene leaders at global AI Safety Institutes & international organizations as well as researchers & policy leaders on the ground in the EU, BRICS, Africa, and beyond for the AI+Policy Symposium. Register now: https://t.co/9yyUQvoNKZ pic.twitter.com/CaQL39O9i0
— Stanford HAI (@StanfordHAI) 12 avril 2024Next week, @StanfordHAI & @StanfordCyber will convene leaders at global AI Safety Institutes & international organizations as well as researchers & policy leaders on the ground in the EU, BRICS, Africa, and beyond for the AI+Policy Symposium. Register now: https://
hai.stanford.edu/events/aipolic
y-symposium-global-stocktaking
… -
AI Interpretability Circuit Discovery Disclosure Concerns Analyzed
By
–
the reason why sholto and trenton avoid answering dwarkesh's question about the highest level circuit discovered by the interpretability team but casually mention circuits for deception and love must be the industry's reluctance to disclose the consciousness circuit
-
Balancing Optimism with Objectivity in AI Development
By
–
Maintaining a positive outlook is important, but it's just as important to recognize when your positivity may be obscuring your objectivity.
-
OpenAI Safety Team Members Behind Altman’s 2023 Ouster
By
–
More info: Leopold Aschenbrenner was part of OpenAI's "superalignment" team and an ally of chief scientist Ilya Sutskever. Pavel Izmailov also spent time on the AI safety team. In case you didn't know, Sutskever was one of the primary figures in the 2023 ousting of Sam Altman
-
OpenAI Fires Researchers for Alleged Information Leak
By
–
Just in from a new report from The Information, @steph_palazzolo
, and @erinkwoo
: OpenAI has terminated researchers Leopold Aschenbrenner and Pavel Izmailov for allegedly leaking information outside of the company. The actual information leaked is currently unknown. -
Devin Implements Safety Controls for Eldritch Website System
By
–
The fact that Devin has implemented both a "Madness Meter" and "ForbiddenKnowledge.js" for the Eldritch website is promising