started · updated
WhatsApp tests AI-powered ‘Scam Alert’ to detect fraudulent messages
WhatsApp is testing a new optional security feature called ‘Scam Alert’ to combat the rise of digital fraud. The tool uses an on-device machine learning model to identify suspicious patterns in messages received from individuals not saved in a user’s contact list.
To maintain end-to-end encryption and user privacy, the AI analysis occurs entirely on the user’s smartphone. Message content is not sent to Meta’s servers for classification, and no automatic reports are generated unless the user explicitly chooses to report a conversation. The model is trained on linguistic signals and conversational structures identified from previously reported scams.
When a potential scam is detected, a warning banner appears within the chat, visible only to the recipient. Users can then choose to block the sender, report the account, or continue the conversation. If the warning is inaccurate, users can mark the chat as ‘trusted’ to prevent future alerts. The feature is currently in a limited beta rollout.
Entities
Claims
What the coverage asserts, and how many sources carry each claim.
- [● 2 SOURCES] Users can block, report, or mark a conversation as trusted if a warning is inaccurate. al-wattan.net · s.alwakeelnews.com
- [● 2 SOURCES] Meta is testing an AI-based feature in WhatsApp to detect fraudulent messages and warn users. al-wattan.net · s.alwakeelnews.com
- [● 2 SOURCES] The AI model for scam detection runs locally on the user's device to maintain privacy. al-wattan.net · s.alwakeelnews.com
- [● 2 SOURCES] Users have the option to share the last five messages with WhatsApp to improve system accuracy. al-wattan.net · s.alwakeelnews.com
- [○ 1 SOURCE] The system scans messages from contacts not in the user's contact list for fraud indicators. al-wattan.net
- [● 2 SOURCES] Fraud warnings are visible only to the user and not to the sender. al-wattan.net · s.alwakeelnews.com