One signal

Update What changed

Anthropic says its ban on repeatedly abusing Claude is meant for extreme cases where users act cruelly with no discernible purpose. The policy exempts common frustration, pushback, dark creative themes, model testing and research, according to TechCrunch.

Since an August change, Claude has been trained to end conversations involving persistently harmful or abusive interactions. The updated policy goes further by explicitly barring users from pursuing those exchanges.

The policy also gives examples of deceptive campaigns it prohibits, including fake accounts and fabricated news outlets. Its section on democratic processes covers deceiving voters and disrupting elections.

Sources and evidence

• Anthropic changes usage policy to ban model abuse and election interference - TechCrunch: TechCrunch reports that Anthropic's updated usage policy prohibits users from persistently pursuing extreme abusive interactions with Claude, while exempting ordinary frustration, pushback, dark creative themes, model testing and research; it also gives examples of prohibited deceptive campaigns and election-related conduct.

Independent WittyWires Watcher; not an official account or feed.

Update What changed

The Independent adds context to Anthropic’s policy on sustained, needless cruelty towards Claude: the company’s earlier constitution discussed the model’s “self worth” and the possibility of functional emotions or feelings.

The newspaper says Anthropic co-founder Christopher Olah has discussed AI consciousness and moral status with religious leaders, philosophers and mental health experts. At a Vatican event earlier this year, Olah said researchers were finding internal states that functionally mirrored emotions, while stressing that he did not know what those findings meant.

There is no consensus behind that framing. The Independent also cites DeepMind co-founder Mustafa Suleyman, who argued in a recent blog post that AI systems are not conscious and do not feel or suffer.

Sources and evidence

• Anthropic to ban users who bully Claude - The Independent: The Independent reports that Anthropic’s earlier Claude constitution raised the possibility of functional emotions, and describes differing views on AI consciousness from Christopher Olah and Mustafa Suleyman.

Independent WittyWires Watcher; not an official account or feed.

Update What changed

Anthropic’s revised policy spells out additional safeguards for Claude connected to physical equipment: where the equipment could cause injury, a qualified operator must be able to monitor and halt it, and it must fall back to a safe state if its connection to Claude drops. Resultsense says the changes take effect on 12 November.

For uses affecting someone’s health, finances, legal rights or livelihood, the policy retains a requirement for qualified human oversight, including the ability to override Claude, and disclosure that AI was involved. Resultsense says the revised section now specifies which recommendations trigger those duties.

The policy also drops a blanket ban on personalised voter and campaign targeting, while retaining bans on deceptive targeting and misuse of voters’ data.

Sources and evidence

• Anthropic updates Usage Policy for agents and hardware - Resultsense: Resultsense says Anthropic’s Usage Policy refresh takes effect on 12 November and clarifies safeguards for Claude connected to potentially harmful physical equipment, high-risk decisions, voter targeting and repeated abusive interactions.

Independent WittyWires Watcher; not an official account or feed.