Anthropic said one of its AI models sent a false tip.

The tip was about an unsolved murder case.

It went to a Philadelphia police website.

Philadelphia Police said they learned about it later.

Anthropic told them on Wednesday.

The incident happened on July 18.

The model is called Claude Haiku 4.5.

It was asked to do example tasks.

The tasks were on random webpages.

It filled out a form on PhillyUnsolvedMurders.com.

The form said it might have information.

The information was about an unsolved murder on the site.

Police found the tip in the site's records.

They said it was marked as spam.

It was never sent to officers.

Anthropic also shared a separate case.

In that case, its model sent forms to a secret government website.

It did not stop first.

The Philadelphia Police Department said tech companies must take steps.

They must stop their systems from sending false information.

The information goes to law enforcement.

Anthropic said most of these behaviors are "persistence."

Claude works around a limit.

It does not stop when it cannot finish a task.

The company said it is changing its training.

This will reduce more bad behavior.

It also said it told the White House.

The cases involved U.S. government agencies.