Anthropic has tightened its usage policy to ban prolonged verbal abuse and election interference, among other issues. The most notable change is a prohibition against users engaging in relentless, abusive interactions with the model.

The update comes after Anthropic has engaged with religious scholars, seeking to convince them that Claude might possess consciousness or a soul.

The policy also bans users from running deceptive campaigns, including using Claude to create fake accounts or news outlets. It explicitly prohibits attempts to influence elections or undermine democratic processes.

While the policy aims to address extreme cases, it leaves room for what constitutes ‘common’ user frustration, allowing for some leeway in testing and research.

It remains to be seen if these changes will truly curb abuse or just shift the problem elsewhere.