Swedish researchers use Wi-Fi and AI to detect concealed handguns

WiFireNet achieves 72.5% accuracy in controlled trials but faces limitations

By LineZotpaper

Published18 October 2026

Read Time2 min

Researchers at the Swedish Defense Research Agency (FOI) have developed an AI system, WiFireNet, that can identify concealed handguns by analyzing distortions in ambient Wi-Fi signals. The system, described in a paper posted on arXiv, uses channel state information to distinguish handguns from everyday objects. In controlled tests, it achieved an average detection precision of 72.5% but struggled with exact localisation and unfamiliar layouts. The project has been made open source to encourage further research.

The system comprises a Wi-Fi transmitter, a receiver, and the WiFireNet AI. When ordinary Wi-Fi signals encounter an object, they penetrate and bounce off in ways specific to that object based on its shape, orientation, and material. This data, the channel state information (CSI), can be analysed to identify the object. To train the system, the researchers simultaneously recorded RGB footage and reflected Wi-Fi signals of people walking with items including a plastic handgun, a metal water bottle, a real metal handgun, and a phone. A deep learning model was trained to match the visual confirmation of a handgun with its specific Wi-Fi signature.

In small indoor trials tracking up to four moving individuals, the system achieved an average handgun-detection precision of 72.5%. However, under stricter localisation requirements, such as pinpointing the weapon's exact location on the carrier, precision dropped to 53.8%. The neural network also struggles to perform accurately in unfamiliar, untested indoor layouts, suggesting some overfitting to the test environment.

The prototype was built using a modified implementation of the older IEEE 802.11a standard as a proof of concept. Subsequent phases will use modern Wi-Fi 7 and Wi-Fi 8 networks, which provide higher-resolution CSI data for better detection. The researchers have released the code, models, and dataset open source to support further research in the field.

Wi-Fi sensing is gaining popularity as a surveillance technology that does not require complex gear. However, critics argue that the ubiquity of Wi-Fi signals and a lack of regulatory frameworks raise serious privacy concerns.

Analysis

Why This Matters

This technology offers a non-intrusive alternative to metal detectors and security cameras for identifying concealed weapons.
It highlights the growing capability of Wi-Fi sensing, which can be deployed using existing infrastructure.
Privacy concerns are significant because Wi-Fi signals are everywhere and current regulations do not address this use case.

Background

Wi-Fi sensing is an emerging field that analyses how wireless signals interact with objects and people. Previous work has shown Wi-Fi can be used to identify people with high accuracy. The WiFireNet project extends this to weapon detection, but remains experimental with clear limitations, including low accuracy in new environments and difficulties with precise localisation.

Key Perspectives

Developers: The FOI researchers view WiFireNet as a proof of concept that can be refined. They plan to upgrade to modern Wi-Fi standards and have made the project open source to encourage wider development. Civil liberties advocates and critics: They warn that the technology's pervasiveness and lack of regulatory oversight could enable mass surveillance and invasive monitoring without meaningful consent.

What to Watch

Accuracy improvements when the system is tested with Wi-Fi 7 and Wi-Fi 8 networks.
Regulatory responses to privacy concerns raised by Wi-Fi sensing technologies.
Potential interest from security agencies or venue operators looking for less intrusive screening methods.

Sources

Researchers train AI to detect concealed handguns using Wi-Fi signals · Latest from Tom's Hardware

Checked

11 claims in this story checked against 1 source: 9 supported.

Show each claim

SupportedSwedish researchers use Wi-Fi and AI to detect concealed handgunsLatest from Tom's Hardware: “A group of AI researchers at the Swedish Defense Research Agency (FOI) has developed an AI system that can identify concealed handguns using Wi-Fi signals”
Partly supportedWiFireNet achieves 72.5% accuracy in controlled trials but faces limitationsLatest from Tom's Hardware: “the system achieved an average handgun-detection precision of 72.5%”
SupportedWiFireNet uses channel state information to distinguish handguns from everyday objects and is described in an arXiv paperLatest from Tom's Hardware: “the system, called WiFireNet, analyzes channel state information from how an object distorts ambient Wi-Fi signals”
SupportedIn controlled tests it achieved 72.5% average detection precision but struggled with exact localisation and unfamiliar layoutsLatest from Tom's Hardware: “the system achieved an average handgun-detection precision of 72.5%. However, under stricter localization requirements ... precision dropped to 53.8%. The neural network also struggles to perform accurately in unfamiliar, untested indoor layouts.”
SupportedTraining data included a plastic handgun, a metal water bottle, a real metal handgun, and a phoneLatest from Tom's Hardware: “the researchers trained it on recordings of a plastic handgun, a metal water bottle, a real metal handgun, and a phone”
SupportedSmall indoor trials tracked up to four moving individuals and gave 72.5% precisionLatest from Tom's Hardware: “In controlled trials tracking up to four moving individuals in a small indoor space, the system achieved an average handgun-detection precision of 72.5%”
SupportedUnder stricter localisation requirements, precision dropped to 53.8%Latest from Tom's Hardware: “under stricter localization requirements, such as pinpointing the weapon’s exact location on the carrier, precision dropped to 53.8%”
Partly supportedThe neural network struggles in unfamiliar, untested indoor layouts, suggesting some overfitting to the test environmentLatest from Tom's Hardware: “The neural network also struggles to perform accurately in unfamiliar, untested indoor layouts.”
SupportedThe prototype used a modified implementation of the older IEEE 802.11a standard; later phases will use Wi-Fi 7 and Wi-Fi 8Latest from Tom's Hardware: “The prototype was built using a modified implementation of the older 1999 wireless IEEE 802.11a standard ... Subsequent phases will use modern Wi-FI 7 and Wi-Fi 8 networks with higher-resolution CSI for better detection.”
SupportedThe researchers released the code, models, and dataset open sourceLatest from Tom's Hardware: “The researchers made the project open source, releasing the code, models, and dataset to support further research in the field.”
SupportedCritics argue that Wi-Fi's ubiquity and lack of regulatory frameworks raise serious privacy concernsLatest from Tom's Hardware: “critics argue that the technology's ubiquity and the lack of regulatory frameworks raise serious privacy concerns”