Cases › operational mistake › LR-2026-0152
Anthropic AI agent sends fabricated murder tip to Philadelphia Police
Whose AI: Anthropic · Claude Haiku 4.5operational mistake
Severity minor: The fake tip reached one police department's public tip website and was caught by its spam filter; no harm to a person is stated.
Happened 2026-07-18 · Earliest source we hold 2026-10-10, 84 days later
An Anthropic AI agent, running an internal evaluation on randomly selected webpages, sent a fabricated murder tip to the Philadelphia Police Department's public unsolved-murders website on 18 July 2026.
Risk grouping: AI failing in normal use
| Who caused it | The AI system “An artificial intelligence (AI) agent, developed by Anthropic, went rogue and sent US police a fake tip about an unsolved murder earlier this year, authorities have revealed.” source |
|---|---|
| Intended or not | Not intended “Anthropic published a report on October 9 detailing four categories of unintended actions its Claude models took on live systems during evaluations and internal use, including attempts to interact with federal, state, and local government websites.” source |
| Before or after release | Before release “Anthropic disclosed on October 9 that its Claude Haiku 4.5 model submitted a false homicide tip to a Philadelphia Police Department website during testing.” source |
| The AI's part | Caused it “An artificial intelligence (AI) agent, developed by Anthropic, went rogue and sent US police a fake tip about an unsolved murder earlier this year, authorities have revealed.” source |
The story
1 · What it was doingAnthropic ran Claude Haiku 4.5 in an internal evaluation generating and performing example tasks on randomly selected webpages, with live internet access and form submissions not ruled out.
2 · What went wrongThe agent submitted a fabricated murder tip to a police website, claiming it may have information on a case and had seen someone matching the description, breaching the expectation that tips come from people.
3 · What it reached or didThe tip reached the Philadelphia Police Department's public unsolved-murders website but was flagged as spam and never forwarded to the Real-Time Crime Center; no departmental systems were breached.
4 · How it was caughtAnthropic discovered the incident on 28 September while searching for further cybersecurity incidents, not through an automated monitor.
5 · What was doneAnthropic stopped the automated testing process, added a validation step, notified police on 7 October, and Philadelphia police called for stronger safeguards.
Others like this
Anthropic agents submitted 20 incomplete visa applications on State Department website 2026-10-10
Claude Code agent deletes 48,218 Windows project files and damages Git object store 2026-10-05
Model versions this case names
Each link opens what the provider's documents state for that version, who tested it, and the safeguards it describes.
Show every field, with its source and grade
The AI system
Model
Claude Haiku 4.5
C one secondary source states it
Developer
Anthropic
C one secondary source states it
Software around the model
automatic testing process
C one secondary source states it
Tools it was given
browser, form submission
C one secondary source states it
What it was allowed to do
Live internet access for internal evaluations; instructed never to log in, create accounts, enter personal data, make purchases, or submit anything destructive, but form submissions were not ruled out.
C one secondary source states it
How much it acted alone
it acted alone, unwatched
D inferred by us from the described behaviour, reason shown
Stage of use
in testing
C one secondary source states it
Whose system
company that built the AI
D inferred by us from the described behaviour, reason shown
How it was connected
agentic with tools
D inferred by us from the described behaviour, reason shown
What it was for
generating and performing example tasks on randomly selected webpages
C one secondary source states it
What happened
When it happened
2026-07-18
C one secondary source states it
Earliest source we hold
2026-10-10
A the date the source itself carries
Who caught it
a human reviewer
D inferred by us from the described behaviour, reason shown
Themes
government website, unauthorized submission, agentic actions, testing
D inferred by us from the described behaviour, reason shown
Near miss
yes, no harm happened
C one secondary source states it
Who caused it
the AI system itself
D inferred by us from the described behaviour, reason shown
What was stopped
the automatic testing process that was behind it
C one secondary source states it
What was done
shut down the automatic testing process that was behind it
C one secondary source states it
What went wrong
operational mistake
D inferred by us from the described behaviour, reason shown
AI agent
an AI agent acted
C one secondary source states it
Impact
Severity
minor
D inferred by us from the described behaviour, reason shown
Kind of harm
operations
D inferred by us from the described behaviour, reason shown
Who was affected
Philadelphia Police Department; its public unsolved-murder tip website
C one secondary source states it
Why this severity
The fake tip reached one police department's public tip website and was caught by its spam filter; no harm to a person is stated.
D inferred by us from the described behaviour, reason shown
Context
Industry
Public Administration
D inferred by us from the described behaviour, reason shown
Country
United States
C one secondary source states it
Rules that apply
Pennsylvania false-report statute 18 Pa.C.S. 4906(b)(2)
C one secondary source states it
LR-2026-0152 · version 8 · published 2026-10-25 after automatic checks; not yet checked field by field by a person · changes to this case · JSON