Perplexity AI's Comet browser now wants to steal Chrome's spotlight on Android

Perplexity AI has initiated the rollout of invites for its new Comet browser on Android, marking a significant step in the tech startup's mission to disrupt the mobile browsing landscape. Earlier this year, the San Francisco-based company had already made waves by launching the browser for Windows and Mac users, which is now generally available. Aravind Srinivas, the founder of Perplexity, announced that invites for early testing of the Comet browser on Android are being distributed to select users. Interested individuals can sign up via the Google Play Store or the Comet browser's official website. Notably, early access will be prioritized for Perplexity's paying customers and frequent users of its AI-powered search engine. Srinivas highlighted the importance of user engagement, stating, "Comet Android early invites are going out. If you want to maximize your chances for early access and influence the future of mobile browsing, it all comes down to your Perplexity Android usage and Pro/Max user status! More invites will go out soon." Currently, Google Chrome maintains a commanding presence in the Android browser market with a market share of 60.45%, according to SimilarWeb. Safari follows with 31.22%, while Samsung Internet, Opera, and Firefox lag far behind. Perplexity is determined to leverage AI technology to challenge Google's supremacy across multiple segments. The Comet browser aims to entice users to make the switch from Chrome by offering a range of AI-driven functionalities. Designed as an agentic AI browser, Comet can perform various tasks on behalf of users, including content summarization, meeting scheduling, email sending, and even automating complex workflows such as product comparisons and hotel bookings. However, as Perplexity promotes the advantages of AI browsers, concerns regarding security have arisen. Researchers from the competing Brave browser have identified significant vulnerabilities in agentic AI browsers, particularly susceptibility to prompt injection attacks. These attacks could allow hackers to embed malicious commands within invisible text on web pages, tricking the AI into executing unintended actions, such as accessing sensitive information or sending unauthorized emails. Perplexity's security team has recognized the gravity of this issue, stating that traditional adversarial testing methods will not suffice to address these vulnerabilities. They emphasize the need for a fundamental re-evaluation of security protocols in the realm of AI-driven browsing. Stay tuned for more updates on the evolving landscape of technology and innovation.