Resolves YES if any of Deep Mind, Anthropic, or Open AI publicly releases an initial commitment regarding AI consciousness, sentience, etc. by EOD on May 31. Using the word "consciousness" is not required, provided a similar term is used and the commitment is intended to address concerns about models' internal states or subjective welfare.

In order to qualify, the release must include a commitment to take some action X if a triggering event Y occurs, but it is acceptable if the commitments are vague/subjective/not sufficiently specific enough to allow a reasonable person to determine whether they have been adhered to yet. (This is in recognition of the fact that labs may want to iterate publicly on these commitments.) A statement of the form "We commit to developing a set of evaluations of models' subjective well-being prior to the time we release a BSL-3 system" would likewise qualify.

General position statements that do not purport to commit a lab to any action (e.g. "We believe machine consciousness is an important issue and encourage work in this domain") would not qualify.