Police investigators misused AI translation tools, including uploading 350,000 lines of Chinese audio to Google Translate
Official police documents released to RNZ show investigators repeatedly breached rules on AI translation tools. One investigator put 350,000 lines of Chinese audio data into Google Translate, an unsecured platform, and others took evidence home to process on personal devices, which a police report said risked exposing confidential information. A six-month test of the Whisper AI system found officers had broken a ban on transcribing English audio many times, and Waikato investigators used AI to generate a document for disclosure in legal proceedings. Whisper was dropped in March 2025 but later re-introduced, and it remains unclear how its use is audited.
What the AI did
Police investigators used AI translation tools on evidence: one put 350,000 lines of Chinese audio data into Google Translate, a public service not secured for police use. Officers also used the Whisper AI system to transcribe English audio, which was banned, and Waikato investigators used AI to generate a document for disclosure in legal proceedings.
First reported October 18, 2026 · Added to the register October 30, 2026 · 1 source
- Deployer
- Police investigators
- Model
- Whisper AI, Google Translate
- When it happened
- Not stated in the sources
- First reported
- October 18, 2026
What this means for you
Could this affect you?
If your staff handle sensitive or confidential data. Any organisation whose staff can paste confidential material into public AI services or take it home to process on personal devices could expose it the same way, as police investigators did when one put 350,000 lines of Chinese audio data into Google Translate.
What to check
- Provide staff with an approved, secure AI tool for sensitive work.
- Enforce clear rules on how staff may use AI tools.
- Audit how staff actually use AI tools.
- Prevent staff from processing confidential material on unsecured public services or personal devices.
Areas of your AI programme this touches
Every fact and its source (7)
- Whisper use discontinuedMarch 2025“Despite this, Whisper was dropped in March 2025”[1]
- AI used for legal disclosure documentWaikato investigators“Waikato investigators not only broke that rule but used AI to generate a document for disclosure in legal proceedings”[1]
- Evidence processed on personal devicesEvidence taken home“taking evidence home to process it on personal devices”[1]
- Police responseWhisper AI re-introduced“The breaches pushed police to re-introduce Whisper AI a year ago.”[1]
- Volume of data uploaded to Google Translate350,000 lines of Chinese audio data“One investigator put 350,000 lines of Chinese audio data into Google Translate, an unsecured platform for police.”[1]
- Risk identified by police reportExposure of confidential information“This risked "exposing confidential information"”[1]
- Evaluation findingNo-English rule broken many times“an evaluation found officers had broken the no-English rule many times”[1]
Sources
- Police AI: Investigator uploads 350,000 lines of audio in risky workaroundrnz.co.nz · October 18, 2026
How this record is classified. Severity M (2/5): harm to individuals, or a policy breach with limited reach. OECD level: hazard, an event that could plausibly have led to harm. Evidence: Confirmed, meaning a company, official or court statement.
The facts in At a glance and in Every fact and its source are quoted from the sources listed. The summary, What the AI did and What this means for you are written from those sources and checked against them automatically before publication. Records reflect what has been disclosed, not everything that has happened. Incidents are found through news feeds and the GDELT Project.
