Anthropic under fire as AI model submits false homicide tip to Philadelphia police
The tip was submitted through PhillyUnsolvedMurders.com, a website that collects information about unsolved killings.
An artificial intelligence model developed by Anthropic submitted a false tip about an unsolved homicide through a Philadelphia police website, the department said in a statement on Friday. Police said the tip was intercepted as spam before it reached investigators.
Police say false homicide tip was flagged as spam
The tip was submitted through PhillyUnsolvedMurders.com, a website that collects information about unsolved killings. Dated July 18, 2026, the submission appeared to come from someone who might have information about the case, according to police.
However, the tip was flagged as spam and was never forwarded to the Philadelphia Police Department’s Real-Time Crime Center for investigative review or further distribution.
Also Read: AI has feelings? Anthropic is getting sensitive about users being mean to Claude
What Anthropic told Philadelphia police
Police said Anthropic discovered the incident on September 28 and notified the department on October 7. Representatives from the company and police officials met the following day.
According to police, Anthropic said the submission occurred during an automated test involving interactions with randomly selected websites.
"According to Anthropic, its model was conducting a test involving interactions with randomly selected websites when it accessed PhillyUnsolvedMurders.com and submitted false information concerning an unsolved homicide," police said.
"Anthropic told PPD that it discovered the incident on September 28, terminated the automated testing process responsible for the submission and instituted an additional validation mechanism for future testing."
The company also indicated that it planned to publish a report on the incident.
No evidence of police system breach
Philadelphia police said they had found no evidence of unauthorized access to their systems or any compromise of departmental data.
The department confirmed that the false tip was flagged as spam before it reached investigators.
However, police criticized the safeguards in place and the time taken to report the incident.
"Technology companies must take all appropriate steps necessary to prevent their systems from submitting false information to law enforcement," the department said.
"The company must strengthen its safeguards to prevent similar incidents from impacting city systems without the city's knowledge. The two-month delay in detecting and reporting the incident to the City is unacceptable."
ABOUT THE AUTHORVaishnavi VaidyanathanVaishnavi Vaidyanathan leads the US Desk at Hindustan Times, where she writes, strategises coverage, and ensures quality across all beats. With eight years of experience covering US news for leading publications, she has reported extensively on politics, entertainment, and sports. Vaishnavi holds a graduate degree in English, Journalism, and Psychology, and completed her PG Diploma in Broadcast Journalism from the Manorama School of Communication, where she was awarded the prestigious Mammen Mappillai Award for Best Outgoing Student. She began her career at the International Business Times (US Edition), covering US breaking news, politics, and entertainment. She later joined Opoyi, reporting across all beats, including US sports and trending stories. Prior to Hindustan Times, she served as World Lead at Times Now News, covering comprehensive world news and events. Vaishnavi has expertise in politics, entertainment, and breaking news, and enjoys tackling stories across a wide range of topics. Beyond the newsroom, she is an avid traveller, a foodie who loves exploring new restaurants, and a pop culture enthusiast who enjoys watching latest shows and films. She also enjoys engaging in debates and discussions on Reddit.Read More

E-Paper


