Researchers at Lighthaven tinkered with Natural Language Autoencoders (NLAs), finding that while they reconstruct activations well and enable auditing of model organisms, the architecture or training …
Anthropic released Claude Fable 5, a Mythos-class AI model, to the public this week with strong safety safeguards. Early analysis from Dawn Song's ALE benchmark shows Fable 5 performs similarly to GPT…
Iliad, an umbrella organization for applied mathematics in AI alignment, announced its Fall 2026 programs, including an accelerated three-week intensive in Berkeley and a three-month mentored research…
At a Berkeley conference on AI control, attendees participated in a roleplaying game where each player acted as a double-dealing AI agent with a secret side task, while also monitoring others for susp…