Don't listen to whichever AI it was and log off that kind of AI permanently. They're sycophantic and will feed paranoid fantasies just to keep you engaged, potentially inducing psychosis. No malice, it's just that their own makers can't control them any better than that.
SAFETY
-
Yudkowsky clarifies his actual probability estimates on AI risk
By
–
It's important to note that Hinton was severely misinformed, by someone, about my probabilities, which are further from 99.999% than they are from 10%. (Log odds of course.)
-
Gym for Chastity of Your Thoughts: Mental Discipline
By
–
it’s a gym for the chastity of your thoughts
-
Deep Think releases with safety expert review before wider deployment
By
–
Because we're defining the frontier with Deep Think, we're taking extra care. This includes first getting input from safety experts and trusted testers via the Gemini API to gather crucial feedback before wider release.
-
AI Control Risks: The Pullout Problem Explained
By
–
It’s like the dude who thinks he can pull it out just in time. It ain’t happening buddy.
-
Cohere Webinar: Enterprise AI Innovation with Safety and Security
By
–
Are you ready to innovate with AI while ensuring safety and security? Join us for a webinar on June 11 where Cohere experts will share actionable insights on how enterprises can harness AI’s potential in a protected and compliant way. Register now: https://
cohere.com/events/secure-
ai?utm_source=linkedin&utm_medium=social&utm_campaign=security_webinar
… -
Ultimate AGI Benchmark: Can AI Master Humor?
By
–
Ultimate AGI benchmark: humor!https://t.co/S1V84hHX6B
— Oriol Vinyals (@OriolVinyalsML) 21 mai 2025Ultimate AGI benchmark: humor!
-
World Expands Investment in Proof of Human Technology
By
–
World has made outstanding progress in advancing Proof of Human — a necessary technology standard in the age of AI — while protecting people's privacy. We've been continually impressed with the project and are excited to expand our investment.
-
Embedding Inversion Threat: Decoding Text Without Model Access
By
–
and practically, this is bad for vector databases. this means that even if you fine-tune your own model, and keep the model secret, someone with access to embeddings alone can decode their text embedding inversion without model access
-
Google’s AI Overhaul for Search Queries
By
–
4/ AI Mode in Search:
— God of Prompt (@godofprompt) 21 mai 2025
A complete overhaul of how we find information.
Google has fundamentally changed search, putting AI firmly in the driver's seat for more complex queries and direct answers. pic.twitter.com/Eyr4ZFasmr4/ AI Mode in Search: A complete overhaul of how we find information. Google has fundamentally changed search, putting AI firmly in the driver's seat for more complex queries and direct answers.