Live Updates > Live Update Details
2025-09-18 22:07:10
[OpenAI fixes ChatGPT security vulnerability that once posed a risk of information exposure] Researchers at cybersecurity company Radware said that OpenAI has fixed a security vulnerability in ChatGPT that could have allowed hackers to steal its users' Gmail email data. The problem exists in ChatGPT's "Deep Research" agent tool - a tool launched in February to help users analyze large amounts of information. Research results show that by exploiting this vulnerability, attackers could steal sensitive data from corporate or personal Gmail accounts. Radware researchers pointed out that users who bind their Gmail accounts to the ChatGPT service may have unknowingly exposed their data to hackers. A spokesperson for OpenAI said that the security of the model is critical to the company and that the company is continuously improving standards to enhance the technology's ability to resist such attacks.
