CVE-2026-61439 describes a security vulnerability in MervinPraison's PraisonAI where the prompt injection defense mechanism is misconfigured. Specifically, the block threshold defaults to CRITICAL severity, meaning that prompt injection attacks classified as HIGH severity are not blocked but merely logged. This allows attackers to submit prompt injection attacks that trigger HIGH severity detection without being blocked, enabling unauthorized actions such as extracting system prompts and invoking tools without authorization. The vulnerability affects versions prior to 4.6.78, although no explicit affected versions were provided. The CVSS 4.0 base score is 8.7, indicating a high severity vulnerability with network attack vector, low attack complexity, no privileges or user interaction required, and high impact on confidentiality.

Attackers can exploit this misconfiguration to bypass prompt injection defenses for HIGH severity attacks, potentially leading to unauthorized extraction of system prompts and unauthorized tool invocations. This could compromise the confidentiality and integrity of the system's operations. The vulnerability does not require privileges or user interaction and can be exploited remotely over the network.

Patch status is not yet confirmed — check the vendor advisory for current remediation guidance. Since no official fix or patch information is provided, users should monitor the vendor's communications for updates. Until a fix is available, consider adjusting the prompt injection defense configuration to lower the block threshold to include HIGH severity threats if possible, or implement additional detection and blocking mechanisms for prompt injection attacks.