Google patents on-device tool to flag misleading messages in real time
Google describes a system that uses a machine learning model stored directly on your phone to check messages for signs of deception. It analyzes the context of the communication to decide if it is trying to mislead you, then sends an alert if it matches a specific category of harmful content. Because the analysis happens entirely on your device, the data does not need to be sent to a server, which helps keep your private information secure.
In the filing’s own words
The systems and methodology described herein provide a technical benefit of allowing for robust real-time detection of communications of a particular category while still maintaining user privacy. In this regard, since the determination is fully conducted on a user's device, there is minimal delay in detection. Moreover, data collected during detection need not be saved locally in a memory device or exported externally, allowing for the maintenance of user privacy. Additionally, the real-time detection of a communication of the particular category itself allows mitigation of harm caused thereby.
What exactly they claim (claim 1)
1. A method comprising: detecting, by one or more processors of a user device, a communication associated with a user of the user device; identifying, by the one or more processors of the user device, contextual information of the communication; determining, by the one or more processors of the user device using a trained machine learning (ML) model stored on the user device, a likelihood value indicative of whether the communication is a communication of a particular category, wherein communications of the particular category are intended to mislead the user of the user device; classifying, by the one or more processors of the user device using the trained ML model, based on a comparison of the likelihood value to a threshold value, the communication as either the communication of the particular category or a communication not of the particular category, wherein the classifying is based at least in part on the contextual information; and upon classification of the communication as the communication of the particular category, generating, by the one or more processors of the user device, an alert to inform the user that the communication is the communication of the particular category.
Facts
- Company
- Google (Google LLC)
- Official title
- SYSTEM AND METHOD FOR IDENTIFYING AND MITIGATING PARTICULAR CATEGORIES OF COMMUNICATIONS
- Publication number
- US20260350152A1
- Application number
- 19198256
- Filed
- May 5, 2025
- Published
- Nov 5, 2026
- Inventors
- 8
- Claims
- 20
Classified under
G06F 21/554 Physics & computingG06F 40/35 Physics & computingG06F 2221/034 Physics & computingMore from Google in Security & privacy
- Google patents privacy score for third-party apps Sep 22, 2026
- Google patents meeting notes that hide names if you ask Nov 10, 2026
- Google patents phone feature to disable unknown trackers Oct 8, 2026
- Google patents camera that deletes images before saving them Nov 5, 2026
- Google patents local face recognition for smart home cameras Sep 10, 2026
Summaries are simplified from the filing’s own abstract and first claim. A patent filing is not a product announcement.
Patent data and drawings: U.S. Patent and Trademark Office (USPTO). Nebula is not affiliated with or endorsed by the USPTO.
