An Oct. 16 investigation found that security flaws in Energize Lab's Eiliko robot service exposed millions of private conversations and could allow outsiders to alter the artificial intelligence responses across roughly 68,000 devices.
Security researcher Sammy Azdoufal said he accessed approximately 13 million chat records while reviewing the company systems that manage the robots' online conversations. The records reportedly contained transcripts, account details, and profiles with personal information. Some photos and audio files were also accessible on the internet.
The Verge reported that more than 31,000 of the profiles belonged to children in the United States. A separate report from Cybernews estimated the number lower, at around 8.3 million conversations, showing that published totals differ. Azdoufal proved he could change a reporter's robot remotely, including modifying its memory and changing how it answered questions.
These findings demonstrate that outsiders have the ability to manipulate the devices, but not that every robot was actually compromised. Follow-up reporting noted that Energize Lab had not provided a meaningful response to questions from The Verge. As of Oct. 19, owners were told to disconnect their robots from Wi-Fi until the company confirms that the issues are resolved.