Florida woman charged with written threats of violence after Claude conversations were reported to police
BIBIBI
AT A GLANCE
Reports said that a Florida woman’s Claude conversations were flagged and reported to police, after which she was arrested and faced charges of making written threats of violence.
Article
【Florida woman used Claude to keep a diary; Anthropic review led to a police report】
Florida woman Carli Michelle Heller of Bonita Springs was arrested for using Claude as a diary. Investigators said that she wrote in September 26 about “shooting up” the Lee County Sheriff’s Office, and said the next day that she had acquired a new gun. Anthropic’s safety system flagged the conversations, and because of their serious nature, the platform referred them to a human review team, which reported them to law enforcement. Police used this information to identify Heller, went to her residence, and detained her. She was charged with making written threats of violence, and the case is scheduled to go to court on November. Anthropic’s consumer terms expressly reserve the right, “in its sole discretion,” to report user inputs, outputs, or conduct to law enforcement. Its privacy policy, effective September 10, 2026, also allows the company to share data with police when it in good faith believes disclosure is reasonably necessary to prevent serious harm. Even if users opt out of having their data used for model training, Anthropic may still use their data for 3 months, rather than 5 years; conversations flagged for review will also be used to improve harmful-content detection capabilities. This is not the first such case. The San Francisco Standard September reported that Anthropic had previously referred a user to police. The user allegedly wrote that they had purchased a AR-15 rifle and placed the company’s CEO Dario Amodei in the crosshairs. The user said it was only a joke and had not been arrested or charged when the report was published. An Anthropic spokesperson said that this was “our safeguards working as intended.”
Key points
01
Reports said that Carli Michelle Heller used Claude as a diary and was arrested as a result.
02
Investigators said that she wrote in September 26 about “shooting up” the Lee County Sheriff’s Office and said the next day that she had acquired a new gun.
03
Reports said that Anthropic’s safety system flagged the conversation, after which a human review team reported it to law enforcement.
04
Reports said that after identifying Heller, police went to her residence and detained her.
05
She was charged with making written threats of violence, and the case is scheduled to go to court on November.
06
The materials said that Anthropic’s consumer terms reserve the right to report user inputs, outputs, or conduct to law enforcement.
07
The materials said that Anthropic’s privacy policy, effective September 10, 2026, allows it to share data with police when it believes disclosure is reasonably necessary to prevent serious harm.
08
The materials said that even if users do not permit their data to be used for model training, Anthropic may still use their data for 3 months; conversations flagged for review will also be used to improve harmful-content detection capabilities.
09
The materials also mentioned another user who allegedly wrote that they had purchased a AR-15 rifle and placed Dario Amodei in the crosshairs; the user said it was only a joke and had not been arrested or charged when the report was published.
10
An Anthropic spokesperson said that this was “our safeguards working as intended.”
AI-assisted interpretation
The following is analysis, separate from reported facts. Verify important claims independently.
This material describes a platform safety-review case:Claude conversations were first flagged by an automated system, then manually reviewed and reported to law enforcement. The material also claims that Anthropic’s terms and privacy policy allow it to disclose data to police under certain circumstances.
Why it matters to readers
This indicates that conversations between users and AI assistants may undergo human review and be provided to law enforcement because of suspected serious threats of violence, and it also reminds users to note the data processing and disclosure circumstances set out in the platform’s terms.
Beginners should regard AI chat as an online service that may be subject to platform safety detection and rules, and should not assume by default that all inputs will be seen only by themselves.
Risks and unknowns
The materials do not include police records or court documents, so the arrest, charges, and hearing schedule cannot yet be independently verified from the original evidence.
The materials do not include the original text of the relevant conversation, so its full context cannot be verified.
The materials do not include Anthropic's consumer terms, privacy policy, or complete statement, so the relevant policy descriptions and scope of application remain to be verified.
The specific processes and scope of data involved in automated flagging, human review, and disclosure to law enforcement cannot be confirmed.
Related Developments
Loading event timeline…
The other user's incident described in the materials, and whether that user was not arrested or charged, lack support from independent documentation.
Related concepts
Human Review
This term is not in the glossary yet. Browse related concepts in the glossary.