Researchers Hack ChatGPT Memories and Web Search Features
Source SecurityWeek
Rated by Threat Radar
Data as of 6 Nov 2025, 15:50 UTC
What to do
AI-suggested
Organizations should monitor vendor communications for patches addressing these vulnerabilities and apply them promptly once available. Until patches are released, restrict access to AI memory and web search features to trusted users and environments. Implement strict input validation and output monitoring to detect anomalous AI behavior indicative of exploitation attempts. Employ network segmentation to isolate AI systems from critical infrastructure and sensitive data repositories. Conduct regular security assessments and penetration testing focusing on AI integration points. Educate users and administrators about the risks associated with AI memory and web search features to reduce inadvertent exposure. Finally, consider deploying AI usage policies that limit the scope of data processed by ChatGPT to minimize sensitive information exposure.
Description
Tenable researchers identified seven vulnerabilities impacting ChatGPT's memory and web search features, including the latest GPT model. These flaws could allow attackers to manipulate or extract sensitive data from the AI's memory or interfere with its web search capabilities. Although no known exploits are currently active in the wild, the vulnerabilities pose a medium-level risk due to potential confidentiality and integrity impacts. Exploitation likely requires interaction with the AI system but may not need elevated privileges. European organizations using ChatGPT or similar AI services that integrate memory and web search features could face data leakage or manipulation risks. Mitigation involves applying vendor patches when available, restricting access to AI memory features, and monitoring AI interactions for anomalous behavior. Countries with high adoption of AI technologies in critical sectors, such as Germany, France, and the UK, are more likely to be affected. Given the medium severity and absence of active exploits, organizations should prioritize awareness and proactive defense measures to prevent exploitation.