Messenger warns that a chat may be a scam and offers an AI review. Should I share the messages, and what can I do next?
The short answer
A warning is a reason to pause. Meta describes an optional AI review that asks to share recent messages from a potentially suspicious new-contact chat, then may suggest blocking or reporting. Read the sharing notice before deciding. An ordinary safety notice and consent to send messages for review are separate things.
What this answer is based onWhat to do
You can stop replying while you decide. Verify the person or organisation through contact details you already trust or find independently. A familiar logo, mutual contact or convincing profile does not settle who is writing.
- Pause the conversation before following links, paying or providing more information. Keep the relevant username, profile link, dates and text privately if you may need to report; avoid redistributing sensitive messages.
- Open Messenger through the app you normally use. Read its actual warning. A screenshot, link or message sent by the contact claiming to be a security check is not a substitute for an in-app notice.
- Identify the choice being offered: a safety warning, an invitation to share recent messages for AI review, or a reporting flow. Read each screen rather than assuming that opening a warning has authorised every later action.
- If AI review is offered, read what that screen says will be shared and how it will be used. Proceed only if you accept that disclosure. If it is unclear or you do not want to share, leave the review flow; do not send extra private information just to test the tool.
- Choose whether to use the available Block or Report controls. Complete the specific action you intend and check its confirmation. Do not assume an AI result submitted a report or blocked the contact for you; follow the options actually shown.
- If there is no alert, no review option or an inconclusive result, independently check the request anyway. Contact the real organisation through its own website or a known contact, rather than the number or link supplied in the suspicious chat. Do not keep a conversation going to make a warning appear.
- If you already gave away a password or login code, use the linked account-security or recovery guide as a separate next step. Preserve the relevant evidence before clearing the chat. An AI review cannot undo information you already disclosed.
The situation
Hypothetical example: A new contact offers a remote job, asks for an urgent deposit and triggers a Messenger warning. You want to stop the pressure and understand the review prompt before sharing conversation content.
The situation and an improvement
Approve every prompt and assume the review has made the contact safe or submitted a report.
Read the sharing choice, verify the request independently and confirm each protective action.
More examples
The warning is missing
A new contact asks for money, but Messenger shows no scam alert.
What to do
Keep the request on hold and verify it through a trusted route. Do not treat the absence of a warning as permission to pay or share details.
A refund helper moves you into a private chat
After you comment on a company's post, someone claiming to be support asks for account details.
What to do
Use that company's independently verified support channel. Do not assume the reply belongs to the company because it appeared below its post.
What this answer is based on
Meta Newsroom · Messenger scam warnings and optional AI reviewabout.fb.comMeta Newsroom · Scam-review test and independent safety checksMeta Newsroom · Messenger Safety Notices and suspicious contactsGuide based on Meta’s materialReview details
Source reviewed:
We read the official source listed for this topic. Other Meta rules are reviewed separately.
Primary source reread: . This does not mean every supplementary source, translation or app screen was rechecked.
Read Meta's public announcements on 16 October 2026. They describe Messenger safety notices, an optional recent-message AI scam review, and a March 2026 expansion to more countries. No account screens, current regional availability, exact settings route, message count, retention or training use were verified. Independent identity checks, evidence preservation and checking completion are ClearRules precautions; detection, enforcement and recovery are not guaranteed.
Material published: March 11, 2026
Include the guide link, language and wording that needs checking. Add an official source link if you have one. Do not send passwords, login codes or photos of identity documents.
Examples are made up to explain the guidance. Meta decides individual account and content cases.
Sources and related topicsPolicy Graph · Demo in development