Anthropic’s September commitment to embed third-party evaluators with ongoing employee-level access, followed by its October 15 announcement granting METR researchers such access and publication rights without veto, set the benchmark that other labs must match by the October 31 deadline. The late-September White House Accord on Super Intelligence, signed by OpenAI, Google, Meta, xAI and others, added voluntary pressure for independent audits, though negotiations continue over publication independence, IP safeguards, and exact access levels. OpenAI has stated it is finalizing contracts with assessors and will share details soon, while Meta and Google show modest market-implied odds around 18% and 14% respectively amid competitive positioning and regulatory scrutiny. With only two weeks left, any new announcements on embedded evaluation partnerships could quickly shift trader sentiment before resolution.