AI Browser Extensions Expose Firms to Data Theft, Akamai Warns

Companies are rapidly adopting generative AI browser extensions and agents to search, summarize and process information inside employees’ web sessions. The convenience broadens the corporate attack surface because tools with access to browsing activity, page content or network connections can potentially expose internal records, customer data and login credentials. Akamai said excessive permissions make these products a growing concern for security teams overseeing enterprise AI use.

Nearly 75% of AI browser extensions reviewed by Akamai requested high or critical levels of access, increasing the risk of sensitive-data leakage and session hijacking. The research also highlighted threats facing AI products including Cursor and Perplexity, such as stolen API keys and prompt-injection attacks. Akamai’s findings underscore the need for companies to restrict extension and agent privileges, approve tools centrally and monitor unusual access patterns.