PlayStation patented an AI account to identify dangerous users

PlayStation has patented an AI system that creates virtual decoy accounts, mimicking the behavior of minors, to identify users attempting to make dangerous contact with them.

PlayStation has patented a Large Language Model Powered Social Integrity System (patent 18/314775) – a virtual account that, with the help of AI, can identify potentially dangerous users.

The system is designed to mimic the behavior of a potential victim and appear as a real account. Specifically, it can imitate the communication style of minors. A suspicious user makes contact with it, after which the system can record their actions even before unwanted communication with a real person.

The AI will look for spam, attempts to obtain personal data, inappropriate messages, cheating, fraud, and attempts to contact minors. Violators will not be automatically blocked, but their accounts will be flagged, and other users will see a warning when interacting with them.

It is currently unknown whether the technology will ever be released. Such developments often do not reach commercial use.