AISN #26: National Institutions for AI Safety, Results From the UK Summit, and New Releases From OpenAI and xAI — LessWrong