Are you ready for an AI assistant that’s both powerful *and* transparent? Perplexity AI just dropped a major upgrade to its Comet Assistant, and it’s all about giving you more control, visibility, and safety. CEO Aravind Srinivas announced the update focuses on making Comet a responsible assistant that operates with user-approved permissions.
This latest version of Comet is engineered to be more like a responsible human assistant – visible, accountable, and careful about independent actions. Let’s dive into the key improvements.
## See What Comet Sees: Step-by-Step Actions
No more black-box AI! Comet now shows you *exactly* what it’s doing in real-time. Whether it’s browsing a webpage, clicking links, scrolling, or following a multi-step process, you’ll have a front-row seat. A new sidecar panel reveals Comet’s reasoning, letting you follow its workflow and understand *why* it takes certain actions. It’s all about clarity and understanding how Comet arrives at its conclusions.
## You’re Always in Control
Worried about AI overstepping? Perplexity has you covered. Comet now asks for permission before taking autonomous actions. When you enter a query, the assistant may ask if it should handle the task. You choose: do it yourself, allow Comet to perform the action once, or grant ongoing permission. This ensures Comet never oversteps and you’re always in the driver’s seat.
## Enhanced Safety for Sensitive Tasks
Comet now pauses when facing high-risk steps like logging into accounts or making purchases, explicitly asking for permission before proceeding. This mirrors the judgment of a trusted human assistant, someone who knows when a decision is too important to take solo. These updates underscore Perplexity’s mission of creating reliable, transparent, and user-centric AI tools across its ecosystem, emphasizing that it operates safely and visibly at all times. It’s about empowering you to decide what you want the assistant to accomplish!




