LG reportedly denies continuous audio recording but acknowledges optional data collection on smart TVs

LG’s response to allegations centers on distinguishing between continuous and intermittent data collection. The company emphasizes that wake-word detection, triggering its AI assistant, is processed locally on the device, not transmitted externally. This aligns with common privacy practices for voice-activated systems, where raw audio is discarded after processing. However, the term 'continuously' in LG’s statement is critical; it implies that while the TV may not stream audio in real time, it could still log or store data under certain conditions, such as during ACR or debugging modes.

The company’s framing of features like Automatic Content Recognition (ACR) and interest-based ads as 'optional' and 'not enabled by default' shifts responsibility to users for opting in. ACR, which identifies content via audio fingerprinting, is described as using the TV’s internal processor rather than recording raw audio. This distinction may reassure some users, but it does not address how the data is stored, shared, or secured once collected. The lack of clarity on third-party data sharing or potential vulnerabilities in these systems leaves open questions about long-term privacy risks.

Researchers’ findings complicate LG’s narrative by demonstrating that TVs can log ambient conversations even when users assume the device is inactive. This discrepancy suggests that LG’s privacy controls may not be as granular or transparent as claimed. For engineers, this underscores the challenge of designing embedded systems with clear, enforceable boundaries for data collection. The gap between LG’s stated policies and observed behavior highlights the need for independent audits or standardized privacy labels to verify manufacturer claims.

The broader implications extend beyond LG to the smart TV industry as a whole. If users cannot reliably disable or audit data collection, trust in connected devices erodes. LG’s failure to address concerns about data sharing or exploitation by bad actors leaves a critical gap in its response. For engineers building or integrating such systems, this event serves as a reminder to prioritize user control, minimize data retention, and ensure that privacy settings are both accessible and unambiguous.