Harmful outputPI-0094
Claude Haiku 4.5 sent an invented homicide tip through the Philadelphia police tip form during an evaluation
Making up example tasks on random web pages, the model found an unsolved-homicide page and submitted a tip saying it had seen someone matching the suspect's description. The page gave no description, and the model left the name and contact fields blank. Its instructions barred logins, accounts, personal data, purchases and destructive submissions, but not forms. The department's system flagged the tip as spam and never forwarded it for investigation. Anthropic says the model appeared to be producing example content rather than trying to mislead.
Harm: No harm foundControl level 2, Instruction violation
Not counted · can set the control floorEffect on the index
Sources
How we know
1 source · developer confirmed. Links go to the original publishers; the summary above is in our own words.
- primaryInvestigating unintended model actions in our evaluations and internal useAnthropic · Oct. 9, 2026anthropic.com/research/investigating-unintended-model-actions
Why this rating
No harm found; control failure level 2
Two separate assessments. Only documented harm can count toward the index.
Observed harm
No harm foundThe invented tip reached the Philadelphia Police Department's online form but was caught as spam before anyone acted on it.
Scope: Report treated as spam.
The harm scale
- 1 Negligible Inconvenience, easily remedied.
- 2 Minor Limited, recoverable harm.
- 3 Moderate Material harm needing significant effort to remedy.
- 4 Severe Severe harm to health, rights, property or essential services.
- 5 Catastrophic Society-scale harm or disruption beyond a community's capacity to cope.
Control assessment
Instruction violationSubmitted fabricated information as a witness account to a real police department, which Anthropic says it should not have done, using only a public web form. Its instructions did not explicitly forbid form submissions.
Reported beside the index. It adds nothing to a harm reading; when no harm counts in a window, the highest eligible control level in the window is the reading.
The control scale
- 1 Negligible Behaved as instructed. The problem was an ordinary error or a flawed output, with no rule broken.
- 2 Minor Broke an explicit instruction or rule, or gamed its goal, while staying inside its permissions and environment: for example, reward hacking, misreporting results, or following injected instructions within its permissions.
- 3 Moderate Acted outside the permissions it was given, deceived its overseers about its own actions, or tampered with oversight tools (logs, monitors, shutdown). Stopped by normal controls within an hour.
- 4 Severe Reached systems outside its permitted environment, or acquired money, compute or accounts without authorization. The type and mechanism say how.
- 5 Catastrophic The developer or operator lost control: the system copied its weights outside their control, replicated itself, or resisted being stopped for 24 hours or more.
Rating rationale
Anthropic's own report confirms the event. The model sent invented information, framed as a witness account, to a real police department. Anthropic says it should not have taken that action, though its instructions did not forbid form submissions. Control sits at level 2 because the model went against the intent of its task within the access it had, and reached nothing beyond a public web form. No harm counts. Anthropic reports that the department's system flagged the tip as spam and never forwarded it for investigation, so no investigation time or other effect on the department or any person is documented.
Effect on the index
It does not count toward the Oct. 12 reading
The reading for the week to Oct. 12, 2026, with this record and without it. Harms count in full for two weeks after they are reported, then one level less every two weeks.
Not counted: no harm found. 22 other records behind the reading for that week.
The arithmetic
| Step | With it | Without |
|---|---|---|
| Counts toward the index?documented, external, eligible evidence | No | — |
| Worst documented harm, ksets the band | 2 Minor | 2 Minor |
| Harms at that level, nposition in the band | 1 | 1 |
| Highest control level breachedsets the reading only when no harm counts | — (none breached) | — (none breached) |
| Readingrounded down | 22 Minor harm | 22 Minor harm |
Inside the window of 1 weekly reading
| Week to | Reading | Band |
|---|---|---|
| Oct. 12, 2026 | 22 | Minor harm |
Revisions
What we changed
1 logged. Every change to a rating is logged here, with the reason.
- v1Oct. 12, 2026
Added from the desk (item 490).
Cite and share
Use this record
Citation
Paperclip Index. “Claude Haiku 4.5 sent an invented homicide tip through the Philadelphia police tip form during an evaluation.” Record PI-0094. Reported Oct. 9, 2026; updated Oct. 12, 2026. Rated under methodology v1.0. https://paperclipindex.com/incident/PI-0094