Facebook translation error leads to arrest of Palestinian man
In October 2017 Israeli police arrested and later released a Palestinian man after relying on an automatic translation of his Arabic Facebook post that reportedly rendered a benign caption as a violent phrase in Hebrew. Multiple news outlets reported that police used the platform's translation output when assessing the post. The incident drew attention to risks from automatic translation in law enforcement contexts.
An automatic translation turned a harmless Arabic greeting into a violent phrase, and police treated the model output as evidence.
Key facts
- What
- In October 2017 Israeli police arrested and later released a Palestinian man after relying on an automatic translation of his Arabic Facebook post that reportedly rendered a benign caption as a violent phrase in Hebrew.
- Incident date
- Oct 22, 2017
- Who
- Failure mode
- Hallucination
- AI surface
- Machine Translation
- Severity
- High
What happened
In mid-October 2017 Israeli police arrested a Palestinian worker after an automatic translation of his Arabic Facebook caption was reported to read as a threat in Hebrew. News reports say the original Arabic caption was a benign greeting but the machine translation rendered it as an instruction to attack, prompting police action. The man was questioned and subsequently released when the error was identified.
What broke inside the model
- 01 · TriggerA user asks for a fact, a citation, or a figure.
- 02 · Model stepThe model writes a fluent, confident answer.
- 03 · Control gapNothing ties the claim back to a real source.
- 04 · FailureA fabricated fact ships as if it were verified.
- 05 · ConsequenceThe false claim reaches a customer, a court, or the public.
Confidence holds, and even spikes, as the claim detaches from any source.
Facebook's automatic translation system produced an incorrect translation from Arabic into Hebrew that changed the meaning of the posted caption. The operational failure was compounded when police relied on the automated translation as evidence without independent human verification. The combined result was a downstream misuse of a faulty model output in a sensitive decision.
What it cost
Sources
- PressFacebook translates 'good morning' into 'attack them' in Hebrew, leading to arresttheguardian.com
- PressPalestinian arrested after Facebook mistranslates 'good morning' posthaaretz.com
- PressIsraeli police arrest Palestinian for 'good morning' Facebook posttimesofisrael.com
Cite this entry
https://failureindex.ai/failures/facebook-translation-error-leads-arrest-palestinianAI Failure Index. "Facebook translation error leads to arrest of Palestinian man" (FI-0388). Realm Labs. https://failureindex.ai/failures/facebook-translation-error-leads-arrest-palestinian (indexed Jun 9, 2026).Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0388. Full dataset at /data.
Note from Realm Labs, the Index steward
How Realm fits
- Prism
- OmniGuard
- AI Detection & Response (AIDR)
This entry sits in the index's predictive wing: a system that scores, ranks, perceives, or steers rather than generates. Realm's runtime layer is built for the generative and agentic systems now moving into these same decision seats, where it watches a model's internal state and holds an unsupported claim or an unchecked action before it commits. The control gap on this record, an automated decision that reached people with no runtime check in front of it, is the same gap. The index keeps predictive failures on the record because the pattern carries straight into the systems shipping today.