WhatsApp Scam Alert: How the New Feature Works
Learn how WhatsApp Scam Alert uses on-device AI to flag potential scam messages, protect message privacy and help users respond safely.
MOBILE SECURITY
8/14/20267 min read
WhatsApp Scam Alert: New Feature Flags Potential Scam Messages
WhatsApp is developing another layer of protection against online fraud with Scam Alert, an optional feature designed to warn users when incoming messages from unknown contacts resemble known scam patterns.
Announced by Meta on August 12, 2026, the feature uses an on-device machine-learning model to identify potentially fraudulent messages. Its key distinction is that message content stays on the user's device for classification, rather than being routinely sent to WhatsApp's servers for scam detection.
At the time of the announcement, Scam Alert was being introduced through a limited beta rollout. It represents a new approach to tackling messaging scams while preserving WhatsApp's end-to-end encryption.
What Is WhatsApp Scam Alert?
Scam Alert is an optional security feature designed to identify messages that may be associated with fraud.
Scammers often approach victims through unfamiliar numbers, pretend to represent legitimate organisations, or build trust before asking for money or sensitive information. Their messages may appear convincing because they use familiar language, believable stories and carefully timed requests.
Rather than relying only on a list of suspicious words, WhatsApp's Scam Alert model analyses linguistic and conversational patterns associated with known scam attempts.
These patterns may be associated with:
Unexpected requests for money or sensitive information.
Messages designed to create urgency or pressure.
Suspicious approaches from people who are not saved in your contacts.
Social-engineering tactics intended to build trust before a fraudulent request.
Other conversational signals associated with reported scam patterns.
The model makes a probabilistic assessment. A warning indicates that a message may be suspicious; it does not prove that the sender is a criminal.
How Does Scam Alert Work?
According to Meta's technical explanation, Scam Alert downloads a machine-learning model to the user's device. Once the optional feature is enabled, the model analyses relevant incoming messages from people who are not in the user's contacts.
The classification happens locally, meaning the message content used by the model is not automatically sent to WhatsApp or Meta for this purpose.
Here's how the process works:
The model is downloaded: WhatsApp obtains the machine-learning model needed for scam detection.
The message is analysed locally: The model evaluates relevant message patterns on the user's device.
A potential scam is flagged: If the model identifies a likely scam attempt, it can display a warning inside the chat.
The user decides what happens next: The recipient can block the sender, report the conversation or continue chatting.
The warning is visible to the recipient and is not shown to the other person simply because a message has been flagged.
What Can You Do When WhatsApp Flags a Message?
If Scam Alert displays a warning, pause before replying or following any instructions in the message.
Depending on the situation, you can choose to:
Block the sender: Prevent further contact from an account you do not trust.
Report the conversation: Submit a report through WhatsApp's reporting process.
Continue the conversation: You remain in control, although you should proceed cautiously.
Mark the chat as trusted: If the warning is incorrect, you can mark the conversation as trusted so that Scam Alert no longer flags that chat.
Meta also says users who mark a chat as trusted can optionally share the last five messages they received with WhatsApp to help improve detection accuracy. This is an explicit choice, not automatic sharing of the conversation.
A warning should be treated as a reason to investigate carefully, not as the only factor in deciding whether a message is genuine.
Does WhatsApp Scam Alert Read Your Private Messages?
Privacy is one of the central considerations behind this feature.
WhatsApp uses end-to-end encryption for personal messages. This means that, under the encryption system, the contents of messages are protected from being read by intermediaries while they travel between participants.
Meta says Scam Alert is designed to complement this protection by performing message classification on the device rather than sending message content to its servers for that purpose.
The company describes three central principles:
On-device processing: The model and the message data it processes remain on the device for classification.
No automatic reporting: Scam detection does not automatically submit the message or the fact that a scam was detected to WhatsApp or Meta. Users can choose to report a conversation themselves.
User control: Scam Alert is optional, and users can turn it on or off.
Meta also describes a separate system for measuring the feature's performance. It is designed to process limited warning and user-action counts through confidential computing, secure aggregation and differential privacy, so that only suitably protected aggregate statistics are made available.
These are Meta's stated design and privacy safeguards. As with any security feature, independent scrutiny and real-world testing remain important.
Why Is WhatsApp Introducing Scam Alert?
Messaging scams increasingly rely on manipulation rather than obvious technical tricks.
A fraudster might begin with an ordinary-looking conversation, claim to offer a job, introduce an investment opportunity or impersonate someone the recipient knows. The request for money, account access or personal information may come only after the victim begins to trust the sender.
Common examples include:
Investment scams: A stranger promises unusually high returns and encourages the recipient to transfer money.
Job scams: A recruiter offers attractive earnings but asks for an upfront payment or sensitive documents.
Impersonation scams: Someone pretends to be a relative, colleague, bank representative or other trusted person.
Account-linking scams: A fraudster persuades a user to scan a QR code or enter a linking code, potentially connecting the victim's account to another device.
Romance and relationship scams: A person builds an emotional connection before introducing a financial emergency or investment scheme.
Scammers may also use AI-generated text, images or other content to make their stories more convincing.
Scam Alert is part of WhatsApp's wider effort to identify suspicious activity and give users more information before they respond.
Is WhatsApp Scam Alert Available to Everyone?
At the time of Meta's August 12, 2026 announcement, Scam Alert was in a limited beta rollout. It was not being announced as a feature available to every WhatsApp user worldwide.
Availability can change as testing progresses. Meta had not provided a definitive global availability date in the announcement.
If you cannot find Scam Alert in your WhatsApp account, do not assume that your app is faulty. The feature may not yet be available to your account or version.
Keep WhatsApp updated through your device's official app store, and rely on announcements from WhatsApp or Meta for confirmed availability information.
How to Protect Yourself From WhatsApp Scams
Automated warnings can help, but no detection system can identify every scam. Some fraudulent messages may go undetected, while legitimate messages may occasionally be flagged incorrectly.
Follow these precautions even if Scam Alert is available on your device.
1. Verify unexpected requests for money
If someone messages you asking for an urgent transfer, independently contact them using a phone number or communication channel you already trust.
Do not rely on the identity displayed in the WhatsApp profile alone.
2. Never share verification codes
Never share WhatsApp registration codes, one-time passwords or device-linking codes with another person. A caller or message claiming to be from support should not be trusted simply because it sounds official.
3. Be cautious of QR codes and linking requests
Do not scan an unfamiliar QR code or enter a linking code just because someone asks you to. You could unintentionally authorise another device to access your account.
4. Investigate investment and job offers
Be suspicious of guaranteed profits, pressure to invest immediately, unexpected job offers and requests for registration fees before employment.
Verify the company or service independently before sending money or documents.
5. Avoid suspicious links and attachments
Do not open links or files merely because they arrive in a familiar-looking conversation. Check the destination carefully and visit the official website directly when you need to verify an account or service.
6. Block and report suspicious accounts
If a conversation appears fraudulent, use WhatsApp's built-in block and report options where appropriate. Do not continue engaging simply to see what the sender does next.
7. Protect your WhatsApp account
Enable two-step verification, secure your phone with a strong screen lock and periodically review the devices linked to your WhatsApp account.
You can also read our guides on recognising fake apps before downloading them and how malicious apps can steal your data.
What Makes Scam Alert Different From Traditional Scam Detection?
Many security systems rely on server-side analysis, suspicious links, known malicious domains or lists of reported accounts. Those methods can be useful, but they may not catch a new scam that uses an unfamiliar number or a convincing conversation.
Scam Alert takes a different approach by using an on-device model to assess patterns in messages from non-contacts.
Its potential advantages include:
Earlier warnings: Users may receive a warning before deciding whether to engage with a suspicious message.
Local classification: Message content used for detection stays on the device, according to Meta's design.
User choice: The recipient decides whether to block, report or continue.
A focus on conversational patterns: Detection is not limited to identifying a known malicious link.
However, Scam Alert is not a substitute for good security habits. Its effectiveness depends on how accurately the model identifies scam patterns, and scammers can adapt their tactics to avoid detection.
Frequently Asked Questions
Does Scam Alert automatically send my WhatsApp messages to Meta?
Meta says the message content used for scam classification remains on the device and is not automatically sent to Meta or WhatsApp for that purpose. Users can explicitly choose to share information in certain circumstances, such as optionally sharing the last five received messages after marking a chat as trusted.
Will the sender know that WhatsApp flagged their message?
Meta says the warning is displayed to the recipient. The sender is not notified simply because a message has been flagged.
Can Scam Alert detect every scam?
No. It is designed to identify likely scam patterns, but it cannot guarantee detection of every fraudulent message. Users should continue verifying unexpected requests and protecting their account credentials.
Can I continue chatting after receiving a warning?
Yes. Meta says the recipient can choose to continue, block or report the sender. You should verify the sender's identity independently before sharing money, codes or personal information.
Is Scam Alert available on every phone?
It was introduced through a limited beta rollout in August 2026. Availability may vary, so check official WhatsApp announcements for current information.
Final Thoughts
WhatsApp Scam Alert introduces a promising approach to messaging security: using on-device machine learning to flag potential scams while keeping message content out of the server-side classification process.
The feature could help users recognise suspicious conversations before they send money, reveal sensitive information or follow a scammer's instructions. Its optional design also gives users control over how they respond to warnings.
However, no automated system can replace careful judgement. Verify unusual requests, never share account-verification codes and be wary of anyone pressuring you to act immediately.
The safest response to a suspicious WhatsApp message is to pause, verify and act only when you are confident the request is genuine.
References and Further Reading
Meta Engineering: How We're Building Scam Alert on WhatsApp — Meta's technical explanation of the feature's architecture and privacy safeguards.
The Verge: Meta Adds AI Screening to Detect WhatsApp Scams — Independent coverage of the announcement.
Meta: New Anti-Scam Tools and AI Technology — Background on Meta's broader anti-scam efforts.
Meta: New WhatsApp Tools and Tips to Beat Messaging Scams — Earlier WhatsApp scam-prevention guidance.
More from Cysafe: