Anthropic Updates Claude Usage Policy to Ban Abuse and Election Interference

Anthropic has updated Claude’s usage policy to prohibit prolonged abuse of its AI models and clarify restrictions on election interference, weapons software and surveillance. The revised usage policy, announced Thursday, October 8, takes effect November 12, 2026.

The most unusual addition prohibits users from repeatedly treating Claude cruelly without any discernible purpose. Anthropic says the restriction applies only to extreme cases and does not cover ordinary frustration, criticism, dark creative content or legitimate AI testing.

The change follows an August 2025 update that allowed Claude to end conversations involving persistent abuse or harmful requests. The latest policy formally prohibits users from deliberately continuing such interactions.

Anthropic has also explored questions surrounding AI consciousness and moral status. As The New York Times reported, the company consulted religious scholars on whether advanced AI systems could be conscious or deserve moral consideration.

Beyond model abuse, the policy prohibits coordinated deceptive campaigns, including networks of fake social media accounts and fabricated news outlets. A section titled “Do Not Undermine Democratic Processes” addresses misleading voters and interfering with elections.

Anthropic also clarified restrictions on weapons-related software and unauthorised surveillance. Most changes formalise or reorganise existing safeguards, while the explicit prohibition on sustained abuse introduces a new rule governing interactions with Claude itself.