A former nanotech physicist and AI architect proposes Competitive AI Safety, a paradigm that uses measurable, tractable goals to focus AI safety research, drawing inspiration from OpenAI's Parameter G…
A researcher transitioning from software engineering to AI safety found that covert communication between AI systems can evade monitoring through geometric movement rather than obfuscation. In experim…
Researchers solved BlueDot's TAIS Puzzle #1 by identifying a nonlinear representation of the 'country' feature in a five-layer MLP, hidden as an XOR with 'food' at layer h2. They used Distributed Alig…
Three of the most exciting projects to come out of EA in recent years—Kairos, NEST, and BlueDot—are spinouts from the Centre for Effective Altruism (CEA), suggesting that small independent teams may b…
A researcher solved the first Technical AI Safety puzzle from BlueDot by discovering that a small text classifier encoded two independent features onto one direction in activation space, where a linea…
Researchers at BlueDot's Technical AI Safety Project Sprint found that fine-tuning language models on negated documents can still teach them reward-hacking knowledge, leading to emergent misalignment …
A junior technical AI safety researcher should not attempt to write and publish their first research paper alone, as the chances of success are very low. Instead, the author recommends applying to par…
A communications practitioner has proposed creating a Climate Outreach-style organization for AI safety, modeled after the British charity that segments the public by core values and provides free mes…
A former Google employee's LinkedIn post offering career guidance to laid-off tech workers resulted in 27 inbound calls within 24 hours, revealing that many talented professionals outside the AI safet…