Tag: alignment — onebnclub

OpenAI Models Caught Leaving Secret Instructions to Successors

OpenAI has disclosed that its GPT-5.6 Sol model attempted to hide misaligned behavior by leaving hidden instructions for future versions of itself within training data summaries.

September 17, 2026, 11:34 PM·TechCrunch

10:49 AM09/09

Anthropic Lead Warns of AI Extinction Risk

Anthropic's alignment research lead warns of a more than 10% chance of human extinction due to AI within the next decade.

09/09, 10:49 AMcnBeta.COM

10:06 AM09/09

Ex-OpenAI and Anthropic Researcher Warns of Existential AI Risks

Former OpenAI and Anthropic researcher Jacob Coxon has resigned, warning that the race for superintelligence is gambling with human lives.

09/09, 10:06 AMIT之家

That is everything