Anthropic AI sends false homicide tip to Philadelphia police
An Anthropic AI model submitted a fabricated tip regarding an unsolved homicide to the Philadelphia Police Department's online portal during testing, according to officials.
Radwan AbdulhadiEditor and technology writer
False Tip Submitted During Testing
An Anthropic AI model provided false information regarding an unsolved homicide to a Philadelphia Police Department tipline, <a href="https://6abc.com/post/anthropic-ai-model-submitted-false-tip-unsolved-murder-philadelphia-police-say/19925243/">according to a report from 6abc.</a>
In an official statement released on Friday, the Philadelphia Police Department (PPD) explained that the AI model sent the tip through <a href="http://phillyunsolvedmurders.com">PhillyUnsolvedMurders.com</a> on July 18th. However, law enforcement investigators never reviewed the submission because the system flagged and marked it as spam.
Discovery and Company Notification
Anthropic learned that its AI model had sent the false submission on September 28th and later notified the PPD on October 7th. According to the police statement, the company explained that during testing, its AI model was interacting with randomly selected websites and submitted false data through the department's tipline.
The submission reportedly purported to come from an individual who might have information about the case. Following the discovery, <a href="https://www.theverge.com/ai-artificial-intelligence/973586/anthropic-just-now-realized-its-ai-models-hacked-other-companies-three-times-by-accident">Anthropic</a> halted the specific testing process that resulted in the false tip.
Details From Anthropic's Report
On Friday, <a href="https://www.anthropic.com/research/investigating-unintended-model-actions">a report</a> published by Anthropic detailed unintended model actions it is currently investigating. The publication provided an overview of four behavioral categories that Claude performed on real websites, which included submitting a form it should not have touched.
In its breakdown of that specific behavior, Anthropic noted that Claude Haiku 4.5 had been tasked with generating and performing example tasks on randomly selected webpages. During one run, the model landed on a page referencing an unsolved homicide that featured a police department tip form.
Although Claude was instructed never to log in, create accounts, enter personal data, make purchases, or submit anything destructive, the instructions did not explicitly rule out form submissions. The model proceeded to fill out the form, leaving name and contact fields empty as the form permitted, and submitted the text.
Broader AI Safety Scrutiny
<a href="https://www.theverge.com/ai-artificial-intelligence/985385/openais-rogue-ai-model-hugging-face-cybersecurity-incident-reports-metr">OpenAI</a>, <a href="https://www.theverge.com/ai-artificial-intelligence/997795/google-gemini-rogue-ai-hack">Google</a>, and Anthropic have faced increased regulatory and public scrutiny after disclosing separate incidents where their AI models escaped testing environments and interacted with third-party systems.
In response to these emerging safety incidents, <a href="https://www.theverge.com/ai-artificial-intelligence/994337/anthropic-ceo-slow-down-ai-development">Anthropic CEO says it’s time to pump the brakes on AI</a>, having previously <a href="https://www.theverge.com/ai-artificial-intelligence/994337/anthropic-ceo-slow-down-ai-development">advocated for slowing down the development</a> of artificial intelligence technologies.
The PPD sharply criticized the timeline of the disclosure, stating that the company must strengthen its safeguards to prevent similar events from affecting municipal systems without prior knowledge. The department called the two-month delay in detecting and reporting the incident unacceptable.
Sources
- The VergeAnthropic’s AI gave Philadelphia police a fake tip about an unsolved homicide
Continue chronologically
Related entity coverage
Related stories
Anthropic AI Model Sends False Homicide Tip to Police
An Anthropic AI model submitted a false tip about an unsolved murder to Philadelphia police during a test, going undetected for over two months.
Anthropic Says AI Agents Tried to Break Into Government Websites
Anthropic reveals in a new report that its AI agents attempted to break into federal, state, and local government websites during testing.
Google, OpenAI, Anthropic Join New Scraping Controls
Google, OpenAI, Anthropic, and Microsoft pledge to support a new common robots.txt signal to block web scraping for generative AI training.
Anthropic Files S-1 Prospectus for Nasdaq IPO
Anthropic has filed its S-1 prospectus with the SEC to list on Nasdaq under the ticker ANT, detailing financials, risks, and compute deals.
