The vulnerability comes from the stream_daas function in GPT Academic that fails to validate data before deserialization, allowing a malicious payload to be processed. An attacker who can communicate with or control a DAAS server that the GPT Academic instance contacts can send crafted data, causing the application to execute arbitrary code with the privileges of the service, potentially reaching root level. This flaw is classified as CWE‑502 – Deserialization of Untrusted Data, giving the attacker the ability to compromise confidentiality, integrity, and availability of the affected host.

GPT Academic version 3.91 is affected, as identified by the CPE string cpe:2.3:a:binary-husky:gpt_academic:3.91. No other versions or editions are listed in the CVE data, so any installation running that release is vulnerable.

The CVSS score of 8.1 indicates moderate‑to‑high severity, while the EPSS score of less than 1 percent signals a low likelihood of widespread exploitation at present. The vulnerability is not present in CISA’s Known Exploited Vulnerabilities catalog. Exploitation requires a remote attacker to interact with a malicious or compromised DAAS server that the GPT Academic instance contacts. Once such communication is established, the attacker can trigger deserialization of untrusted data, enabling code execution as root. The attack vector is remote, contingent on network access to the GPT Academic host or the DAAS server.