← Timeline Sep 18 2026 AI Evaluator Forum publishes minimum conditions for embedded safety reviewers SAFETYPOLICY Original reportingAaievaluatorforum The letter calls for independent editorial control, privileged access, direct board communication and protection from retaliation. Signatories act in personal capacities; the letter does not establish binding lab commitments. Around the same timeSep 19 2026MONEYOPENNaive AI reaches a $1.42B valuation seven months from founding, pre-releaseBThe InformationSep 18 2026ADOPTPOLICYSingapore confirms Waymo’s planned path to commercial rides in 2028AltaSep 18 2026SAFETYResearchers disclose reaching OpenAI's internal monorepo via a forum bug chain, with Claude writing much of the exploitAhacktronSep 18 2026POLICYSAFETYCalifornia accelerates independent AI oversight and requests shutdown-safeguard proposalsAgovSep 18 2026SAFETYMONEYAnthropic names Accenture as an embedded evaluatorAnewsroomSep 18 2026POLICYINFRAVirginia bars state executive-branch secrecy agreements for data-center projectsBCardinal News