Samsung Smart TV User Tracking Exposed: Always-On Microphone and AI Surveillance Uncovered

Cybersecurity Researcher Uncovers Extensive Samsung Smart TV User Tracking

A cybersecurity researcher known as log4j on X has exposed what they call a "privacy horror" after reverse engineering their Samsung Smart TV firmware1

1

. The investigation uncovered a file called 'security-policy.xml' listing multiple data collection functions including an always-on microphone, automatic content recognition, geo-positioning, network mapping, targeted advertising services, and user profiling capabilities. The discovery has sparked widespread privacy concerns about how much information modern smart TVs collect from users without their explicit awareness.

The investigation began when the researcher noticed unusually personalized advertisements appearing on a Samsung TV purchased in 20252

2

. After exploiting previously unknown vulnerabilities to obtain root access to the device running the Tizen operating system, the researcher reportedly uncovered an extensive collection of monitoring capabilities. These included audio categorization, screen-content recognition, and cross-device advertising identifiers that enable sophisticated smart TV user tracking.

AI-Driven Surveillance Through Convolutional Neural Networks

Perhaps most concerning is the discovery of a region-specific convolutional neural network running directly on the television itself2

2

. This AI-driven surveillance system allegedly analyzes viewing habits including which videos users watch, their playback progress, and which advertisements they skip. The researcher speculates that this extensive background processing could contribute to the sluggish responsiveness frequently experienced with modern smart TVs, though no performance measurements have been provided to substantiate this theory.

The firmware analysis also revealed that Samsung TVs without built-in microphones can identify microphones available through connected devices such as soundbars and Bluetooth speakers1

1

. This capability potentially allows audio collection even without a microphone integrated into the television itself, expanding the scope of data collection beyond the TV hardware.

How Data Collection Keeps TV Prices Low

Jake Moore, Global Cybersecurity Advisor at cybersecurity firm ESET, suggested that extensive data collection might explain why TV prices haven't increased significantly in recent decades1

1

. "TVs haven't really increased in price like some products over the last two decades and that may be because customer data collection is keeping those costs down," Moore explained. This business model essentially subsidizes hardware costs through the monetization of user data for targeted advertising purposes.

James Bore, a cybersecurity and technology consultant, emphasized that while these findings aren't entirely new, they add crucial awareness to something most people don't realize. "The amount of monitoring that smart devices, from TVs to voice assistants to cars, do isn't secret as such, just not something promoted," Bore noted1

1

. He pointed out that privacy policies and End-User License Agreements typically disclose these practices, but most people don't read those documents.

Legitimate Features Versus Hidden Surveillance

A community note on the researcher's post highlighted that Samsung openly advertises some of these features1

1

. The company's Pet Care and Family Care tools on Vision AI TVs listen for sounds like barking or crying and alert owners through the SmartThings app. Similarly, Samsung's automatic content recognition technology can identify viewing content and collect viewing-history information, though Samsung maintains that the feature requires user consent and can be disabled.

However, independent verification attempts have yielded mixed results. Hugging Face engineer Xuan-Son Nguyen conducted a separate experiment using a Flipper Zero and observed the television activating its microphone for approximately five seconds before stopping2

2

. Nguyen stressed that the results were inconclusive, particularly because the television had already been rooted and several settings were disabled, making it difficult to determine normal operating behavior.

Regulatory Scrutiny and Privacy Implications

Smart TV tracking has already attracted regulatory attention. In February 2026, Samsung reached an agreement with the Texas Attorney General requiring explicit consumer consent before collecting or processing automatic content recognition viewing data from Texas consumers2

2

. This regulatory action demonstrates growing concern about how companies monitor viewing habits and collect personal data without transparent disclosure.

Moore emphasized that privacy concerns extend beyond Samsung devices. "The concern is how much of our data is being collected and who it's being shared with," he said1

1

. "It is possible to go through TV settings to minimise the amount of data access points are being used and analysed, but saying that, privacy shouldn't be something you have to hunt through dozens of settings to find."

AI-Powered Reverse Engineering Trend

Moore predicted that this discovery could be the start of a wave of similar findings as AI-powered reverse engineering becomes more accessible1

1

. "AI-assisted reverse engineering may make it easier and faster for researchers to uncover and explain how devices collect, process, and share data," Bore agreed. This democratization of reverse engineering tools means more hidden functionality in everyday devices is likely to surface in coming months.

The investigation also resulted in an unexpected confrontation with OpenAI. The researcher reported having their OpenAI account suspended for alleged "Cyber Abuse" after using its tools during the reverse engineering process2

2

. They disputed the suspension, maintaining that the tools were used for legitimate local firmware analysis rather than malicious exploitation.

What Users Should Watch For

Bore warned that the public should be more concerned about how "blasé" we've become about accepting companies "constantly profiling us to manipulate us into spending more money on things we don't need"1

1

. Some security-conscious individuals are already changing their behavior. Muneeb Ali, co-founder of Stacks, a Bitcoin layer-2 network, revealed he doesn't handle private crypto keys in rooms with devices, stating "Only the paranoid survive."

Last week, smart TV owners started using AI coding agents like Claude Code to strip ads and tracking software from their sets1

1

. One open-source project, Cleanyourtv, now lists tested setups for 107 TV models from 37 brands, though it only fully works on Android-based TVs and not Samsung's Tizen operating system. The researcher's findings remain preliminary, with a comprehensive technical write-up reportedly in preparation. Until detailed evidence and reproducible testing become available, users should remain vigilant about the data their smart devices collect and take steps to minimize unnecessary data sharing through available privacy settings.