CVE-2025-65720
CriticalCVSS 9.8Exploitation Probability (EPSS)
Low risk50th percentile - higher than 50% of all known CVEs
Summary
An issue in GPT Researcher v3.3.7 allows attackers to execute arbitrary commands on a victim system via user interaction with a crafted HTML page.
Risk Assessment
The risk for the organization includes full compromise of the victim system, potentially leading to data theft, malware installation, or further propagation of the attack within the network.
Recommendation
Immediately update GPT Researcher to the latest version that fixes this vulnerability. Until updated, avoid opening suspicious HTML pages in environments running the vulnerable version.
Other vulnerabilities in GPT Researcher
See all- CVE-2026-37006Critical
Vulnerability in the WebSocket endpoint of gpt-researcher v0.14.7 and before. An unauthenticated remote attacker can achieve code execution via malicious Model Context Protocol configurations.
- CVE-2026-5633High
Vulnerability in gpt-researcher up to version 3.4.3 in the ws Endpoint component. Manipulation of the source_urls argument can lead to server-side request forgery (SSRF). The attack is remote and the exploit has been publicly disclosed.
- CVE-2026-5632High
In gpt-researcher up to version 3.4.3, a missing authentication vulnerability was found in an unknown function of the HTTP REST API endpoint. The attack can be performed remotely and the exploit has been made public. The project was informed but has not responded.
- CVE-2026-5631High
Vulnerability in gpt-researcher up to version 3.4.3. The extract_command_data function in backend/server/server_utils.py in the WebSocket component allows remote code injection via manipulation of the args argument. The exploit has been publicly disclosed.
- CVE-2026-5630Medium
A cross-site scripting vulnerability has been found in assafelovic gpt-researcher up to version 3.4.3 in the Report API component in file backend/server/app.py.
- CVE-2026-5625Medium
gpt-researcher up to 3.4.3 has an XSS vulnerability in the WebSocket interface via the 'task' argument in researcher.php. Remote exploitation is possible and the exploit is public.
Original NVD description (English source)
An issue in Open Source GPT Researcher v3.3.7 allows attackers to execute arbitrary commands on a victim system via user interaction with a crafted HTML page.