Deep-TEMPEST: deep learning eavesdrops on HDMI screens via EM emanations
heyshrutimishra · x · 2026-11-10
A 2024 paper from Universidad de la República (arXiv:2407.09717) shows deep learning reconstructing screen content from the unintended electromagnetic emissions of HDMI cables and connectors — a digital-era upgrade to classic TEMPEST attacks.
- The challenge: unlike analog VGA, HDMI's 10-bit encoding brings much larger bandwidth and a nonlinear mapping between signal and pixel intensity, so analog-era eavesdropping rigs yield blurry, unreadable images;
- The approach: recast the problem as an inverse problem and train a deep learning module to map the captured EM signal back to the displayed image; signal analysis determines the tuning frequency, and simulated samples replace a real TEMPEST setup for training;
- Results: average character error rate on text improves by over 60 percentage points vs. prior implementations;
- The paper and code are public, with related projects like tempestsdr also linked.
Related event: Deep-TEMPEST: Deep Learning Reconstructs Screens from HDMI Leakage(2 posts)→
More from Safety
- Anthropic-backed MAEM paper reads an AI's inner thoughts when it learns it will be deleted — belindazli · 2026-11-10
- Bloomington, Indiana passes ordinance banning superintelligence within city limits — yoavgo · 2026-11-10
- Claude Code Security Audit Uncovers Baby Monitor Streamed on Public Site — mikeflache · 2026-11-10
- Meta and Sierra Lead Working Group on AI Agent Identity Standards — mikeflache · 2026-11-10
- WEF Briefing Paper: Governing Agentic AI Systemic Risks — mikeflache · 2026-11-10
- Fable 5.2 system card shows blocked kernel dev on certain accelerators, reportedly Huawei and DeepSeek — casper_hansen_ · 2026-11-10