Loading story…
Researchers at the Hong Kong University of Science and Technology (Guangzhou) and Hong Kong Polytechnic University disclosed InjectEave, an active electromagnetic side-channel technique that recovers audio played through wired and wireless headphones from up to 30 meters away, including through concrete walls. Instead of relying on weak passive emissions, an attacker transmits a tuned RF carrier that nonlinear components—such as amplifiers, ADCs, and power converters—mix with low-frequency audio signals, causing the audio to be re-emitted on a measurable carrier; the demonstrations used commercially available SDR equipment.
The researchers also observed injection-induced leakage from smart-home power consumption and analog sensor inputs. In a closed-loop scenario involving a Flyingvoice VoIP landline, they combined eavesdropping with AI voice cloning and synthesized-audio injection, demonstrating both confidentiality and integrity risks. The findings identify analog circuitry and connected-device wiring as an attack surface that can bypass protections aimed at digital communications and may not be fully addressed by conventional cryptography or ordinary EM shielding.
Track how attackers are adapting to this technology.
2 events from the most recent confirmed update back to the earliest known activity.
Haoran Yan, Ziyu Shao, Shuhao Zhang, Qinhong Jiang, and Yan Long submitted a paper introducing Injection-Induced EM Side Channels and the InjectEave attack. The research demonstrated RF-injection-assisted recovery of headphone audio at up to 30 meters, including through-wall scenarios, and described smart-device and landline-phone leakage and manipulation risks.
The paper "Injected and Leaked: Actively Inducing Side-Channel Leakage Using Electromagnetic Injection and Hardware Nonlinearity" was accepted at USENIX Security '26. The reported work tested the attack against commercial headphones, a Flyingvoice VoIP landline, and smart devices, including a closed-loop eavesdrop-synthesize-inject demonstration against the landline.
Follow how adversaries are adapting to this technology, and where it touches your stack today.
3 references tracked. Mallory keeps watching after this page renders.
cryptika.com
Open sourcecybersecuritynews.com
Open sourcearxiv.org
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Researchers at Texas Tech University have developed a prototype system that enhances voice authentication by combining traditional audio verification with physical measurements of jaw and cheek movements. This inertial-speech verification approach uses helmet-mounted motion sensors to create a unique motion profile for each speaker, making it significantly harder for attackers to impersonate someone using only voice deepfakes. The system addresses the growing threat of generative AI-powered voice cloning, which has already been used in scams and fraudulent approvals, by requiring attackers to also replicate the physical act of speaking, not just the sound. Separately, researchers at the University of California, Irvine, have demonstrated a theoretical attack called "Mic-E-Mouse," which leverages the high sensitivity of modern optical mouse sensors to capture minute vibrations caused by nearby conversations. Under specific conditions—using mice with extremely high DPI and polling rates—adversaries could potentially eavesdrop on speech in secure environments by intercepting mouse data. Both studies highlight the evolving risks and countermeasures in the realm of audio-based attacks, emphasizing the need for multi-layered defenses as sensor technology and generative AI continue to advance.
Jul 12, 2026
Researchers disclosed three vulnerabilities in Bluetooth audio devices built on Airoha systems-on-chip that can allow an attacker within radio range to gain unauthorized access to device memory and communications. The flaws, tracked as **`CVE-2025-20700`**, **`CVE-2025-20701`**, and **`CVE-2025-20702`**, stem from missing authentication in GATT and Bluetooth BR/EDR components and weak permission controls in a proprietary protocol, affecting headphones and earbuds sold by brands including **Sony, Bose, Jabra, JBL, Marshall, Beyerdynamic, and Teufel**. ERNW researchers said the bugs can be chained to hijack the Bluetooth link between a phone and an audio device, issue commands to the paired phone through the **Hands-Free Profile**, eavesdrop on conversations, and potentially overwrite firmware for remote code execution on the accessory. The attack requires proximity because exploitation is limited by Bluetooth range, but defenders were urged to apply vendor firmware updates and investigate or restart devices showing abnormal Bluetooth behavior.
Jul 24, 2026
Researchers at the University of California, Irvine, have demonstrated a novel side-channel attack called Mic-E-Mouse, which leverages the high-performance optical sensors in modern computer mice to eavesdrop on nearby conversations. The attack exploits the extreme sensitivity of these sensors, which can reach up to 20,000 dots per inch (DPI), allowing them to detect minute vibrations on a desk surface caused by sound waves from human speech. By capturing these subtle vibrations, the mouse can effectively act as a makeshift microphone, even though it was never designed for audio input. The research team developed a sophisticated pipeline that combines advanced signal processing and machine learning to filter out background noise and reconstruct intelligible speech from the raw movement data collected by the mouse. In their experiments, the researchers achieved a speech recognition accuracy of up to 61% using a standard $35 optical mouse, demonstrating that the attack is feasible with inexpensive, widely available hardware. The process resulted in a significant improvement in signal quality, with up to +19 dB gain, making the reconstructed audio much clearer. This attack is classified as a side-channel attack because it gathers information through an unintended physical pathway, rather than exploiting software vulnerabilities or network weaknesses. Traditional security measures such as antivirus software and firewalls are largely ineffective against this type of threat, as the attack relies on physical properties of the hardware and does not necessarily leave traces in system logs or trigger conventional security alerts. However, the attack does require the installation of specialized software on the target computer to collect and transmit the mouse sensor data, which could potentially be detected by security tools. The need to exfiltrate data from the compromised system also increases the risk of detection by network monitoring solutions. The researchers published their findings on arXiv and highlighted the broader implications for hardware security, emphasizing that even trusted peripherals like computer mice can become vectors for sophisticated espionage. The discovery underscores the importance of considering physical and side-channel threats in security assessments, especially in environments where sensitive conversations occur near computers. While the attack is currently a proof of concept, it raises concerns about the potential for real-world exploitation, particularly in high-security settings. Organizations are advised to be aware of such unconventional attack vectors and to monitor for suspicious software that could facilitate side-channel data collection. The research also suggests that future hardware designs may need to account for such risks to prevent similar vulnerabilities.
Jul 12, 2026