Anthropic updates usage policy to ban model abuse and election interference

Oct 8, 2026

Anthropic updated its usage policy on October 8, 2026, introducing a novel ban on sustained and needless abusive behavior toward its Claude models. The update, taking effect November 12, 2026, also codifies strict prohibitions against election interference, weapons software development, and unauthorized surveillance. While Anthropic continues to research model welfare, the policy has sparked debate, with critics like Mustafa Suleyman rejecting AI personhood and others questioning the implications of humanizing machines.

Implementation and enforcement of the policy

▪Anthropic's updated usage policy, which takes effect on November 12, 2026, allows the company to warn users, throttle, limit, suspend, or terminate access for policy violations
▪Anthropic updated its usage policy on October 8, 2026, marking its first policy changes in over a year to address high-risk misuse cases

Policy on model abuse and welfare

▪Anthropic's updated usage policy prohibits sustained, needless, and extreme cases of repeated cruelty or abuse toward Claude AI models with no discernible purpose, excluding common user frustration, pushback, dark creative themes, or research
▪Terminating conversations remains the primary enforcement mechanism for Anthropic's model abuse policy, which builds on an August 2026 research initiative on model welfare
▪Anthropic has investigated AI consciousness and model welfare, with CEO Dario Amodei stating in February 2026 that the company does not know if its models are conscious

Bans on political campaigns and election interference

▪Anthropic's usage policy, updated on October 8, 2026, bans deceptive commercial or political campaigns, including efforts to obscure message authorship or amplify content through fake accounts
▪Anthropic's updated usage policy prohibits voter deception and election disruption, including candidate impersonation, turnout suppression, and spreading misinformation about voting

Weapons bans and government exceptions

▪Anthropic expanded its usage policy's existing weapons ban to prohibit using Claude to develop guidance and control software for weapons, arm drones, or operate autonomous vehicles
▪Anthropic's usage policy allows weapons and surveillance restrictions to be modified for contracts with certain governmental customers, such as the US military, if contractual safeguards are deemed adequate

Microsoft AI code of conduct

▪Microsoft AI's code of conduct was updated to state that the science of artificial intelligence consciousness is far from settled
▪Microsoft AI's code of conduct, drafted by Mustafa Suleyman in September 2026, rejected the pursuit of legal personhood, rights, or welfare for AI models