New AI Vulnerabilities Enable Data Leakage and Exploitation in ChatGPT

Researchers have identified multiple vulnerabilities in OpenAI's ChatGPT, including seven critical flaws that allow attackers to exfiltrate personal data, inject malicious prompts, and manipulate the AI's memory. The vulnerabilities, collectively termed 'HackedGPT', affect user privacy and security, with implications for both individual and enterprise users. Additionally, a side-channel attack named 'Whisper Leak' has been disclosed, enabling eavesdroppers to infer chat topics even under encryption.