A 30-year-old woman from Florida arrested after sending threats to the Claude chatbot

As TechSpot reported, citing the arrest report, 30-year-old Carli Michelle Heller from Bonita Springs, Florida, allegedly wrote in a conversation with Claude on September 26 that she intended to attack the sheriff's office.

According to investigators, the chatbot's security systems allegedly flagged the content, after which it was passed on to a human for review. The reviewer found it a credible threat and notified authorities.

Lee County Sheriff's Office officers identified the author of the entries and went to her home. The arrest took place without incident.

Woman: Claude was her "diary"

Heller allegedly told investigators that she used Anthropic's tool as a kind of diary. Reports indicate that she denied any actual intention to carry out the attack.

She was charged with making a written threat of violence. Under Florida law, this type of act can be classified as a second-degree crime if the message contains a threat to kill, injure another person, carry out a mass shooting or act of terrorism.

The court will decide about the woman's guilt.

Anthropic allows for an exception in life-threatening situations

Anthropic declares that it generally does not disclose user data at the request of authorities without an appropriate legal basis, such as a warrant or subpoena. However, its policy provides an exception for emergencies.

“We may make an exception if we believe that there is an immediate threat that could result in imminent physical harm or death and that prompt communication of information would prevent it,” the company says in its published policy.

The case is another reminder that conversations with chatbots do not have to have the confidential status that users might ascribe to a personal diary or private correspondence.