AI research and discussion on LessWrong. www.lesswrong.com
Analyses China's AI safety regulation, including the non-binding Framework 3.0 on loss-of-control risks, CAC model filing gaps, content-focused enforcement, and a court opinion on developer liability.
Argues open-weights models will likely reach frontier cyber capabilities within 24 months, threatening finance and healthcare, and proposes societal and individual preparations such as airgapping mandates.
Argues that open-weight models undermine the defenses proposed for near-future AI risks, since guardrails can be fine-tuned away, use cannot be monitored, and bad actors have access by default.
Builds cross-layer transcoder features for Qwen3-0.6B and 1.7B base models, uses topological data analysis to group features, and reports more abstract features than prior work, without a safety focus.