Observed facts: Anthropic published the update on October 8, 2026, and says it takes effect November 12. It added a prohibition on sustained, needless abusive or cruel behavior toward models, but explicitly excludes common frustration, pushback, dark creative themes, testing, and research.
Mechanism: The policy sets the boundary at repeated cruelty without discernible purpose, while Claude’s ability to end persistently abusive conversations remains the primary enforcement mechanism. Anthropic says the addition aligns with an existing capability to end rare conversations on Claude.ai and Claude Code.
Why now: Anthropic says it refreshes the policy annually in response to model capabilities and customer feedback; it also reports that Claude has taken on longer, more independent work. The update is effective November 12, 2026.
Implication: The rule formalizes a limit on user conduct without claiming that routine criticism or testing is prohibited. It also places this boundary within a broader policy refresh addressing deceptive activity, elections, weapons, surveillance, and high-risk deployments.
Evidence boundary: The source is Anthropic’s own policy announcement. It describes the intended scope and enforcement approach, but does not provide independent evidence about user behavior, enforcement frequency, or outcomes.
Watch: Whether enforcement remains focused on ending conversations, and how Anthropic applies the stated distinction between purposeless cruelty and legitimate testing or critique.
New branch: No prior New Runtime record was supplied, so this begins a distinct analytical record of the published policy update rather than asserting a link to an earlier item.