LG rejects surveillance claims after smart TV network and audio tests
LG rejects surveillance claims after smart TV network and audio tests
LG's smart televisions became the subject of a privacy dispute in September 2026 after Gamers Nexus investigated their network activity and voice features. Notebookcheck's report on 7 September described tests involving local-device discovery, microphone recordings and information retained while a television was disconnected from the internet.
LG rejected the surveillance allegations. In an official statement dated 12 September, the company said its televisions do not continuously record conversations and that voice recognition, viewing recognition and personalised advertising are optional features. The disagreement concerns both what the tested hardware could do and how those capabilities operate in normal use.
What the investigation reported
According to Notebookcheck, Gamers Nexus worked with Level1Techs and security researchers on retail LG OLED televisions, including the G5. Their network captures showed discovery of other devices, such as phones and smartwatches, alongside information about nearby wireless networks. The reporting also described voice prompts appearing in readable text logs, raising questions about what a household television retains beyond the immediate command.
The researchers reportedly demonstrated microphone capture while a screen appeared to be off, with locally stored material sent after connectivity returned. Notebookcheck also reported security vulnerabilities undergoing responsible disclosure. These were accounts of the investigators' tests, rather than an independent replication by the publication. They should not be read as proof that every LG television, with every combination of settings, behaves identically.
LG's explanation
LG's response distinguished listening for a wake word from an activated voice session. It said wake-word processing happens on the television and does not produce stored or transmitted conversations when no trigger is detected. During an active session, however, nearby speech may enter the recognition results. The company denied retaining voice data for later upload after a network connection returns.
LG also said discovering compatible devices serves connectivity functions, not advertising profiles. It described Automatic Content Recognition, or ACR, as a separate system for identifying watched content through fingerprints. According to the company, ACR and advertising require consent, and ambient or voice audio is not supplied to LG Ad Solutions for advertising. These are LG's explanations of its systems, not independent findings.
The disagreement did not end with the statement
What Hi-Fi? reported on 15 September that Gamers Nexus had challenged LG's response in a follow-up video, accusing it of contradictions. The publication also highlighted a crucial distinction in the dispute: some demonstrations concerned what someone could do after compromising a television. That is a security question as well as a privacy question, and it differs from establishing what an unmodified set routinely sends to its manufacturer.
A demonstration of a device's microphone capability cannot, by itself, establish who receives recordings during ordinary use. Equally, a company's description of intended behaviour does not independently settle whether a particular software version or setting matches that description. The September reports leave those evidence boundaries important for readers assessing the claims.
Why the UK and US context matters
Earlier academic work provides a narrower, independently researched picture of viewing-data collection. A study released on 10 September 2024 by Gianluca Anselmi and colleagues examined Samsung and LG televisions in the UK and US. It found ACR network traffic when sets displayed conventional television or content supplied over HDMI. Using a separate playback device therefore did not necessarily prevent viewing recognition while the television itself remained connected.
In those experiments, opting out stopped traffic to ACR servers, whereas account login status did not determine whether that traffic occurred. The researchers also found differences between UK and US behaviour, including the servers contacted. This was research into viewing recognition on the tested products, not confirmation of the later microphone allegations. Its practical relevance is that the input cable, consent settings and regional implementation all matter; simply describing a television as a screen does not establish what its connected software is doing.