Hark Pro launches with free access and a privacy-policy training exception

Free access and paid plans for Hark Pro

Hark Pro received a wide release on October 6, 2026, offering free access and paid subscriptions for heavier use. Paid plans start at $20 per month, according to Fast Company.

Hark describes an artificial intelligence assistant that can use connected services and operate a browser to complete tasks with user permission. The stated purpose is to do work on a user’s behalf, rather than only provide answers.

That promise comes with a clear limit: Hark’s own policy warns that its AI can make mistakes. It asks users to review both the assistant’s outputs and its actions.

Visible browser activity is part of the trust pitch

Hark Pro displays the assistant’s web navigation in a small window, TechCrunch reports. Design lead Abidur Chowdhury presents that feature as a way to build user trust.

The company’s launch messaging also emphasizes privacy and describes personalization as focused on helping users finish tasks. Those are company representations, not independently verified privacy guarantees.

Seeing browser activity and understanding data use are separate issues. The navigation window shows what the assistant is doing online; the privacy policy describes how certain content may be used afterward.

Flagged content can be used for model training

Hark’s privacy policy says inputs or outputs flagged for possible usage-policy violations can be used to train its internal safety-classification and generative models. These models classify potential safety issues and generate content. The provision concerns flagged material, not a stated practice of training on every interaction.

The policy says that material is separated from user IDs for training, but may be linked back to users for enforcement. Removing the direct user link therefore does not mean it can never be restored.

For prospective users, the practical distinction is between permission to act, how content is handled, and whether the work is correct. Hark describes permission-based task execution, but its policy still allows this training use and calls for users to review the results.