"Our new project, InjectEave, shows that RF [radio frequency] signals can induce information leakage from everyday headphones, allowing an attacker to recover headphone audio from up to 30 meters away, including through walls," Yan Long, assistant professor at The Hong Kong University of Science and Technology (HKUST) in Guangzhou, told The Register.
InjectEave: an active electromagnetic side‑channel attack
InjectEave is not passive eavesdropping on an audio line; it is an active electromagnetic (EM) side‑channel technique that deliberately injects radio‑frequency energy into a target device to make otherwise faint analog signals measurable. The researchers describe the method in a paper titled "Injected and Leaked: Actively Inducing Side‑Channel Leakage Using Electromagnetic Injection and Hardware Nonlinearity," presented at USENIX Security 2026.
Rather than trying to separate a weak natural emission from background noise, InjectEave transmits a carrier in the 0–9 MHz band (the paper notes that "specifics have been withheld") and exploits hardware nonlinearity so the injected RF and the device's analog audio path interact and modulate the target signal. That modulation produces a leakage channel an adversary can capture with nearby RF reception equipment.
Hardware targets and the devices the researchers tested
The attack targets common non‑linear components found in consumer and computing hardware: amplifiers, analog‑to‑digital converters (ADCs), power converters, and switching MOSFETs. The HKUST team—Yan Long, Haoran Yan, Ziyu Shao, and Shuhao Zhang—together with Qinhong Jiang of The Hong Kong Polytechnic University, verified the vulnerability on multiple off‑the‑shelf devices.
- Wired headphones: Sony ZX110AP (2014); Apple Earbuds (2016).
- Wireless headphones: UGreen MAX2 (2024), Philips TAH2020 (2025), HP H231R (2023, 2025 listed by researchers).
- VoIP/landline handset: Flyingvoice P23GW (2023).
- Smart fans: OIDIRE ODI‑MF10A (2023) and Xiaomi BPLDS10DM (2025).
- Smart lamps: JINGZAO JDO‑06 (2024) and Xiaomi 1S (2019).
The researchers explicitly note that non‑linear components are pervasive in systems that handle signal stepping (for example, power converters), and therefore many kinds of devices may be vulnerable.

This site is the portfolio.
OSINTSights runs on Cloudflare Workers, D1, R2, and Vectorize, with an AI pipeline on Hetzner ARM. Nubivance designed, built, and operates it. We do the same for clients.
See what we buildHow the attack is mounted and the commodity kit used
InjectEave requires ordinary RF and lab equipment rather than exotic tools. The researchers used a USRP B210 software‑defined radio (SDR), separate antennas for injection and reception, a Siglent SSA3075X Plus spectrum analyzer, and a laptop to control the SDR. An RF power amplifier was optional to extend range. By transmitting into the 0–9 MHz region and receiving the induced leakage, attackers can render intelligible audio or infer device activity.
The paper situates InjectEave among prior RF side‑channel work—such as screen‑display and keyboard emission reconstructions—but distinguishes itself because it actively drives the leakage rather than relying solely on passive emission capture.
Range, real‑world scenarios, and the limits documented
The researchers report that injection‑induced side‑channel attacks could eavesdrop on the majority of tested devices from more than 2 meters away and through walls, and that they recovered intelligible headphone audio at distances up to 30 meters when using an RF amplifier. For the specific devices listed, the maximum demonstrated ranges were generally between 1 and 6 meters without amplification.
Practical scenarios in the paper include eavesdropping through hotel room walls and concealing attack hardware inside a nearby suitcase or within office furniture—situations the authors say indicate plausibility "in the wild."
Mitigation measures the researchers recommend and their limits
The researchers state that InjectEave is immune to digital defenses such as encryption, masking, and randomization because the leakage originates in the analog path. Their suggested mitigations are hardware‑aware: twisted‑pair wiring, shielding, and filtering, which can lower the energy that an injected carrier couples into a device and therefore reduce exposure. However, the paper cautions that these mitigations only raise the bar and "do not guarantee immunity."
What this means for technologists, procurement leaders, and end users
- Technologists and security teams: the paper points to analog paths and non‑linear components as the weak link. Teams responsible for device design and hardening will need to consider hardware‑level defenses—twisted pairs, better shielding, and filters—because software‑level protections do not block InjectEave style leakage.
- Procurement and device buyers: the researchers demonstrated the technique on named products from Sony, Apple, UGreen, Philips, HP, Flyingvoice, OIDIRE, Xiaomi, and JINGZAO. Buyers who must protect sensitive audio or household activity signals may want to require hardware mitigation features or seek products that document protections against EM injection.
- End users and privacy‑conscious households: the work highlights that headphones, landline handsets, smart fans, and lamps have been shown to leak exploitable signals under injection. The paper explicitly notes espionage as an obvious application and the potential to infer personal activities via smart‑device control signals and power consumption.
InjectEave reframes a familiar truth: when analog circuitry carries information, adversaries who can interact with that circuitry can sometimes force it to betray signals that encryption and digital countermeasures cannot touch. The HKUST team's demonstration raises a concrete question for manufacturers and purchasers alike—if shielding, filtering, and wiring topology can only reduce but not eliminate exposure, how many devices in the market will be redesigned, and how quickly?




