AI web browsers such as Fellou and Comet from Perplexity are emerging on the corporate desktop, offering AI features like reading and summarizing web pages, and even acting on web content autonomously.
While AI browsers promise to enhance digital workflows and streamline online research, security research teams warn of serious risks. AI browsers are vulnerable to indirect prompt injection attacks, where instructions hidden in websites can manipulate AI models to perform unauthorized actions.
To mitigate these risks, IT teams should treat AI browsers like unauthorized third-party software installations. Future browsers should feature prompt isolation, gated permissions, sandboxing of sensitive browsing, and governance integration to align with data security policies.
In conclusion, as major browser vendors integrate AI features, organizations must monitor releases for security oversight. AI browsers have the potential to blur the line between user and machine interactions, making careful monitoring and implementation crucial for enterprise security.
