LG smart TVs have come under scrutiny after an investigation raised concerns that the devices could monitor audio and identify other devices connected to a home network, even when the television appeared to be in standby mode.

The investigation by Gamers Nexus, conducted with security researchers MrBruh and uturn, examined LG televisions through packet capture and firmware analysis.

It claimed the TVs could record audio, log data and scan local networks for connected devices such as phones and printers.

Researchers also reported that some voice-related data could be stored on the television when it was offline and retrieved later.

The investigation further raised concerns about the way LG TVs sample audio and video inputs to identify content being watched.

LG has strongly rejected the claims, saying they are “not true” and that its televisions do not collect or record ambient conversations unless voice functionality has been activated by the user.

In a statement to Tom’s Hardware, LG said voice data is processed only when the voice button on the remote is pressed and held, or when a wake word such as “Hi LG” is recognised after the Far-Field voice recognition feature has been enabled.

The company said that when the television is in standby mode, it only monitors for the wake word if the feature has previously been activated.

If the wake word is not detected, the audio is processed locally and immediately deleted without being transmitted to its servers, LG emphasised.

The electronics giant confirmed that its TVs can scan for nearby devices on the same network, but said this is a standard function used by smart TVs and other smart home devices.

The company also said its Automatic Content Recognition (ACR) feature is opt-in and is used for personalised recommendations, services and advertising, adding that ACR data is not used for advertising when users have not provided consent.

LG also did not address all of the vulnerabilities and other claims raised in the investigation, including allegations concerning stored transcripts.