Anthropic's Opus 5.5 hits 2.3x compute efficiency, reshaping AI research timelines
Anthropic's Opus 5.5 model achieves 2.3x compute efficiency in P-Zero Research's TasteVal benchmark, accelerating AI research timelines.
Recursive Self-Improvement
Latest Recursive Self-Improvement coverage.
OpenAI partners with 3rd-party assessors to audit frontier AI safety
OpenAI enlists independent assessors to expand third-party evaluations of frontier AI safety, outlining new principles and priorities.
OpenAI calls for US-led global standards on recursive AI development
OpenAI urges US to lead international effort on standards for frontier AI, including recursive self-improvement
Anthropic's Claude now leads 26% of its own model R&D, raising recursive self-improvement questions
Anthropic reveals Claude AI model now handles 26% of its successor's development, sparking debate on AI safety and control.
Jacob Coxon criticizes Anthropic's 'paranoid' RSI push, warns of existential risk
Former Anthropic researcher Jacob Coxon slams the company's recursive self-improvement strategy and expresses concerns about AI safety.
Thinking Machines' Schulman: RSI still needs an outer loop
John Schulman of Thinking Machines discusses the challenges of recursive self-improvement in AI, emphasizing the need for an outer loop.
Anthropic and OpenAI researchers warn of 10%+ chance of AI extinction risk by 2030
Anthropic and OpenAI researchers warn of existential risks from advanced AI models, calling for a slowdown in development.
Anthropic's AI cuts alignment research costs to $4/hour, outperforms humans
Anthropic's Automated Alignment Researcher (AAR) reduces alignment research costs to $4/hour, outperforming human methods in six hours.
Anthropic's AI system outperforms human researchers in alignment tasks for $4/hour
Anthropic's automated system improves AI alignment benchmarks faster and cheaper than human researchers, per a new paper.
OpenAI ex-researcher warns of 50% valuation drop, starts RL dataset firm with $700k equity locked
Andrew Ho, ex-OpenAI researcher, warns of potential 50% valuation drop, starts RL dataset firm with $700k equity locked until IPO
Recursive Superintelligence secures $410M AWS compute deal, prioritizing agent count over staff
Recursive Superintelligence signs a $410M multiyear compute deal with AWS, focusing on agent count over headcount for AI development.