← Back to Tech & Science

OpenAI Patches Critical ChatGPT Flaw Allowing Code Execution via Shared Links

Tech & ScienceAI-Generated & Algorithmically Scored·

AI-generated from multiple sources. Verify before acting on this reporting.

SAN FRANCISCO (AP) — OpenAI has deployed an emergency patch to address a critical security vulnerability in its ChatGPT platform that allowed malicious actors to execute unauthorized code within user sessions and access private conversation data through compromised shared links.

The flaw, classified as a P1 severity incident by the company, was remediated on Oct. 31, 2026. Security firm Parameter identified the vulnerability, which exploited a mechanism in how ChatGPT processes shared conversation threads. The defect enabled attackers to inject and run arbitrary code when a victim clicked on a maliciously crafted link, potentially granting full access to the user's session history and sensitive information.

OpenAI confirmed that the vulnerability was actively being exploited in the wild prior to the patch. The company stated that the fix immediately neutralized the threat, preventing further unauthorized access or data exfiltration. Parameter researchers noted that the attack vector required no direct interaction with OpenAI's backend infrastructure, relying instead on the client-side rendering of shared chat content.

The incident highlights the risks associated with the rapid expansion of generative AI features, particularly those involving user-generated content sharing. By allowing users to share specific conversation threads via URL, the platform created a potential entry point for attackers to bypass standard authentication measures. Once a victim opened a compromised link, the injected code could run within the browser environment, effectively hijacking the session.

OpenAI has not disclosed the number of users affected or the duration of the vulnerability's existence before detection. The company also declined to comment on whether any data was successfully stolen during the window of exposure. Security experts warn that while the immediate threat has been neutralized, users who clicked on suspicious links prior to the patch may still be at risk if their credentials or session tokens were captured.

Parameter advised all ChatGPT users to review their account activity and change passwords immediately if they suspect they interacted with a malicious link. The firm also recommended that organizations using enterprise versions of the platform conduct internal audits to ensure no unauthorized access occurred within their networks.

OpenAI stated it is working with global cybersecurity agencies to investigate the scope of the breach and identify the threat actors responsible. The company has pledged to implement additional security controls, including stricter validation of shared link parameters and enhanced monitoring for anomalous code execution patterns.

Questions remain regarding the origin of the exploit and whether similar vulnerabilities exist in other AI-driven platforms that utilize shared conversation features. As artificial intelligence systems become more integrated into daily workflows, the balance between user convenience and robust security remains a critical challenge for developers and regulators alike.

Discussion

0 / 2000