CVE-2026-23866

WhatsApp AI rich responses could trigger arbitrary URL processing

Facebook · Meta AI · WhatsApp rich responses for Instagram Reels

First documented disclosure · Review checkpoint

Summary

A user could make another device process media from an arbitrary URL.

Severity and exploitation evidence

Medium · 4.3/10 · CVSS 3.1. Severity source.

Exploitation not established: Meta says it has not seen evidence of exploitation in the wild.

Exploitation evidence source

Technical severity is not a measure of observed attacks or of risk to your deployment. Attempts, demonstrations, and successful compromise retain the qualifications in the source.

How the weakness works

AI rich-response messages inadequately validate the source of a communication channel.

Conditions required

An attacker with the vendor’s low-privilege prerequisite sends a crafted rich response; the vendor CVSS vector requires no victim interaction.

Potential impact

Unintended media fetching and invocation of operating-system-controlled custom URL handlers; arbitrary code execution is not established.

Affected versions and configurations

WhatsApp Android 2.25.8.0 to before 2.26.7.10; WhatsApp iOS 2.25.8.0 to before 2.26.15.72, per the structured affected ranges.

Fix and mitigation status

Android 2.26.7.10 and iOS 2.26.15.72.

Practical next steps

Update the affected WhatsApp application to its corrected platform build.

Operating lesson

Treat URLs embedded in AI-rendered rich content as untrusted external input.

Evidence and limits

Historical backfill discovered through the newly reviewed November WhatsApp advisory index. The full Meta CNA and original advisory establish May 1 publication. The narrative uses “to” at the endpoint, while the structured affected ranges explicitly use lessThan; the exclusive fixed boundaries are retained. Functional exploit maturity in the CVSS temporal vector is not evidence of malicious exploitation. CNA publication is 2026-05-01; this is distinct from the earliest reviewed public source and correction dates. Meta assessments: CVSS 3.1 4.3.

Sources