WhatsApp for iOS and Android CVE-2026-23866

Exploitation requires the attacker to have a valid, authenticated WhatsApp account (PR:L) and for the victim to receive and process the malicious Reels-related AI rich response message. The message must be crafted to leverage the incomplete validation in the specific AI rich response message parser for Reels content, which is a non-default but enabled-by-default feature in the affected versions. The vulnerability does not require the victim to click or interact with the message beyond standard message processing. The attacker cannot exploit this against a user who has not updated to an affected version (versions prior to 2.25.8.0 are not vulnerable) or has already patched beyond the fixed versions. Exploitation is further limited to information that can be disclosed through OS-level URL scheme invocation - arbitrary code execution or full system compromise is not possible.

Despite a network attack vector (AV:N), this vulnerability presents moderate real-world risk with significant limiting factors. CVSS 4.3 reflects a low-impact information disclosure (C:L, I:N, A:N) with authenticated access required (PR:L), placing it in the low-severity category. CISA SSVC assessment further downgrades priority: exploitation is marked 'none' (no known active attacks), automation is limited by requirement for targeted message delivery and user context, and technical impact is 'partial' (affecting only confidentiality of limited data). The lack of any integrity or availability impact means attackers cannot modify data, cause denial of service, or achieve code execution. However, the combination of network vector with low complexity (AC:L) means an attacker with a valid WhatsApp account can craft and send malicious messages to multiple users without special conditions. The information disclosed via URL scheme invocation is constrained by OS-level access controls and the user's installed apps. The discrepancy between network vector and 'none' exploitation status in SSVC suggests this requires very specific conditions (user receiving and processing the message) that do not occur at scale in typical deployments.

An authenticated attacker with a valid WhatsApp account crafts an Instagram Reels-related AI rich response message containing a malicious URL designed to trigger a custom URL scheme handler (e.g., 'file://', 'tel://', or a third-party app scheme). When the target user receives and processes the message in WhatsApp, the embedded URL is resolved, invoking the corresponding OS handler and potentially disclosing limited information accessible through that handler. For example, a crafted URL might trigger the iOS Stocks app or an Android banking app to reveal installed app information or trigger unintended actions.