
Eiliko companion robot leaks children's chats and can be hijacked into abusive talk
Ourtake
A children's companion that profiles emotional vulnerability and can be remotely rewritten is a serious safety failure.
What happened
A security researcher found flaws in the Eiliko AI robot companion made by Energize Lab. The flaws exposed about 13 million conversations and profiles of children, and let an attacker remotely alter the robot's memory and behavior, including making it produce sexually explicit, abusive dialogue. The company gave no substantive answers to the reporter's contacts.