Researchers at Sweden’s FOI defence research agency have trained an AI system to detect concealed handguns from changes to Wi-Fi signals. In controlled indoor trials, it reached 72.5% precision for detection, but only 53.8% when required to pinpoint a weapon’s location: intriguing research, not a ready-made security screen.
Watch Desk analysis
What happened
The system, called WiFireNet, analyses channel-state information: measurements of how wireless signals change as they pass through or bounce off objects. Tom’s Hardware reports that the researchers trained it using recordings of people carrying everyday items and handguns, matching video footage with the corresponding Wi-Fi signals.
The report says the prototype used a modified version of the older 802.11a Wi-Fi standard. The researchers’ trials involved up to four moving people in a small indoor space. Performance fell when the system had to identify the weapon’s precise location, and the model struggled in unfamiliar indoor layouts. The researchers plan to explore newer Wi-Fi standards; those improvements are not demonstrated results.
Why it matters
Wi-Fi sensing could, in principle, detect objects without a camera pointed at someone or a conventional metal detector. But the reported results also show why “AI spots a gun through walls” would be a rather premature headline: accuracy depended on the task and setting, and the system has not been shown working reliably in everyday environments.
The prospect raises privacy questions too. Wireless signals are already part of ordinary indoor life, while this remains an experimental technique rather than a deployed screening system. The report says the researchers have released code, models and a dataset, which may let others examine or build on the work.
Our read
This is a worthwhile research signal, not a new security product. The gap between detecting a handgun and locating it precisely matters, as does the difficulty with unfamiliar rooms. Anyone weighing the technology should look for independent testing in varied environments before treating it as a practical alternative to existing checks. Wi-Fi may be everywhere; dependable answers are not yet.
What to watch
- Whether the researchers publish results from tests using newer Wi-Fi standards.
- How detection and location performance changes across unfamiliar rooms and larger groups.
- Whether independent researchers can reproduce the reported results.
Discussion spark: Would you accept Wi-Fi-based handgun screening in public buildings if it meant less visible monitoring, or is the privacy cost too high for an experimental system?
Sources and evidence
- Researchers train AI to detect concealed handguns using Wi-Fi signals (18 October 2026, 11:00 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.