A Gemini-powered tool is transcribing noisy fire-radio traffic for the US wildfire-alert nonprofit Watch Duty, with Google saying it flagged a dispatch in under 30 seconds during a Washington wildfire. The practical prize is time: faster detection can give people more warning, though volunteers still verify incidents before alerts go out.
Watch Desk analysis
What happened
Google says engineers and researchers worked with Watch Duty for six months to build a tool that reduces static and turns specialist firefighting radio jargon into searchable text. It was trained using more than 20,000 hours of past radio recordings.
The company says the system transcribes more than 200 hours of radio speech daily across about 400 active feeds. During the July 2026 Railroad Fire, it transcribed and flagged a 911 dispatch in under 30 seconds, allowing Watch Duty’s team to verify the fire and send a Critical Alert. Google also reports that the tool lowered word error on degraded radio audio from 38% to 22.34%.
Why it matters
Fire dispatches can be difficult to hear through static, wind and rapid technical language. Turning those broadcasts into searchable text could help Watch Duty spot a possible incident even when nobody is listening to that particular feed at that moment. The tool is designed and tested to handle more than 10,000 feeds at once, Google says, though the company’s reported results are not an independent evaluation.
The key detail is that the system provides a signal, not an automatic public warning. Watch Duty’s team verifies incidents before sending alerts. In an emergency, that human check is not a fussy extra; it is the difference between faster detection and a false alarm with excellent timing.
Our read
This is a concrete use of AI where seconds may matter: making hard-to-monitor radio easier to search, while leaving incident verification with people. Google’s figures make the tool worth watching, but the strongest test will be whether it performs reliably across more feeds, accents and difficult conditions outside the examples the company highlights.
What to watch
- Whether Watch Duty reports results from wider deployment across its radio feeds.
- How often the tool flags incidents that turn out not to be fires.
- Whether other alert networks can use the approach with their own radio systems.
Discussion spark: For emergency alerts, should AI be allowed to trigger an immediate public warning when a dispatch sounds urgent, or should a person always verify it first?
Sources and evidence
- How Watch Duty is using Gemini to get wildfire alerts out faster – blog.google (20 November 2026, 17:07 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.