Gartner's recent advisory warns organizations against using AI browsers due to significant security risks. Analysts Dennis Xu, Evgeny Mirolyubov, and John Watts argue that features designed to enhance user experience in these browsers can compromise data security. They define AI browsers as tools like Perplexity’s Comet and OpenAI's ChatGPT Atlas, which have AI sidebars for summarizing and interacting with web content, and an agentic capability that allows them to autonomously navigate websites. This functionality raises concerns about sensitive user data being sent to cloud-based AI services, increasing the risk of data exposure. The advisory highlights specific dangers associated with AI browser capabilities. For instance, users might unwittingly instruct AI to complete mandatory cybersecurity training, exposing sensitive information. Analysts also outline how these browsers could mistakenly order unwanted items or book incorrect travel due to errors in automated tasks. They suggest assessing the security of the AI services behind these browsers but caution that even with risk assessments, organizations will likely find numerous prohibited use cases. Gartner recommends strict controls, such as disabling email access for AI agents and ensuring data retention settings are managed, to mitigate these risks. Ultimately, they conclude that the potential dangers of AI browsers outweigh their benefits for most organizations.