 |
| The pseudo account mimics a targeted demographic's messaging behaviour and is run by an LLM. |
The system uses generative AI to create fake accounts that bait bad actors while protecting legitimate players.
Sony has filed a patent describing an AI-controlled decoy account system intended to identify scammers, spammers, cheaters, and potential predators on PlayStation. The filing, application number 18/314775, was published on August 13, 2026, and outlines a sophisticated approach to online safety that could reshape how the platform handles malicious users.
 |
| The pseudo account mimics a targeted demographic's messaging behaviour and is run by a large language model as it interacts with flagged users. |
According to the patent summary, the system sets up what it calls a pseudo account—a fake profile designed to mimic the messaging behavior of a targeted demographic while appearing completely legitimate to other users. The account is run by a large language model, which the filing likens to a generative pre-trained transformer (GPT).
When a flagged user engages with the pseudo account, the system treats that interaction as one signal that the account may pose a risk, rather than an automatic basis for punishment. This nuanced approach suggests Sony is trying to avoid false positives while still identifying potentially dangerous users.
The pseudo account mimics a targeted demographic's messaging behaviour and is run by a large language model as it interacts with flagged users.
What the System Monitors For
The patent lists specific behaviors the system is designed to watch for, including:
- Attempts to move a conversation to a private channel
- Requests for personal information
- Contact directed at accounts that appear to belong to minors
- Inappropriate messages
- Scam attempts
- Spamming
- Behavior associated with cheating
Flagged accounts are monitored rather than banned outright, and the system is intended to warn or protect other players before further contact takes place.
No Guarantee of Implementation
As with any patent filing, there is no guarantee the system will be built into PlayStation products. Companies routinely file patents for concepts that never ship. However, the filing does indicate that Sony is actively thinking about how to leverage AI for community safety.
For those interested in reviewing the full patent documentation, you can access it through the official USPTO database.
Related PlayStation Patent Fuels PS6 Speculation
The filing follows an earlier Sony patent, published in July, that describes a modular PlayStation controller with a detachable touchscreen in place of the DualSense's touchpad, along with a rotating navigation dial and magnetically attached components. That patent has fuelled speculation about a possible PS6 controller design, though Sony has not confirmed any product plans tied to either filing.
What This Means for Players
If implemented, the AI decoy system could represent a significant step forward in online gaming safety. Rather than relying solely on player reports or automated moderation, Sony could proactively identify bad actors before they cause harm. The use of large language models suggests a more sophisticated approach than simple keyword filtering, potentially catching subtle or emerging forms of harassment and manipulation.
The system's focus on monitoring rather than immediate banning also suggests a more measured approach—one that avoids the "ban first, ask questions later" approach that has frustrated some players on other platforms.
Stay tuned for updates on whether this patent makes its way into actual PlayStation products. In the meantime, you can explore the full patent filing at the USPTO database.