Anthropic publishes a revised Responsible Scaling Policy by March 31st
Why I called it
Anthropic's of November 2nd 2026 says its estimates for future models indicate it 'may imminently meet' the policy's definition of dramatically accelerated progress attributable to AI, and that it is 'considering updates' to the policy to keep humans in control of development. A company that has told the a threshold is near, while preparing to sell stock, has every reason to show investors and regulators what it will do when the line is crossed. What would make this wrong: the IPO process pushes policy changes past the quiet period, or the company decides the existing text already covers the case.
The call, in full. By March 31st 2027, Anthropic publishes a new version of its , dated after November 3rd 2026, on its own website.
Scoring criterion. RESOLVES CORRECT if, between November 3rd 2026 and March 31st 2027 23:59 UTC, Anthropic publishes on anthropic.com a Responsible Scaling Policy document carrying a version number or effective date later than November 3rd 2026. RESOLVES WRONG if no such revised policy document is public by that time. Capability or risk reports, blog posts and amendments that describe possible changes without publishing a revised policy do not count.
The criterion is the machine-checkable version: a prediction that cannot be settled by a third party against a public source fails the build before it reaches this page.