Gartner warns of significant cybersecurity risks with AI browsers

As outlined in The Cyber Express, Gartner has issued a strong advisory to organizations regarding the adoption of AI browsers, cautioning that these innovative tools present substantial cybersecurity risks that make them unsuitable for general use.

The advisory highlights that AI browsers' autonomous web navigation and transaction capabilities can bypass traditional security controls, leading to sensitive data leakage, erroneous transactions and credential abuse. Default settings prioritizing user experience can exacerbate these risks, as active web content, browsing history and open tabs may be sent to cloud-based AI back ends. While AI sidebars are susceptible to prompt injection, Gartner's primary concern lies with the agentic capabilities of browsers like Perplexity's Comet and OpenAI's ChatGPT Atlas. These features allow the AI to act autonomously, increasing the potential for phishing attacks, inaccurate reasoning leading to errors and the loss or abuse of credentials. Sensitive data viewed on these browsers could be transmitted to cloud services, posing a significant risk.

Gartner strongly recommends organizations block all AI browsers for the foreseeable future due to these identified cybersecurity threats and the nascent nature of the technology. For organizations with a higher risk tolerance, limited experimentation with tightly controlled, low-risk use cases is permissible, provided robust security measures and minimal sensitive data exposure are maintained.

Source: The Cyber Express