User files contain confidential and high-stakes client information. We appreciate the security and privacy needs legal documents require. jhana does not ever use, train AI models with, share or sell any uploaded files. Files are converted into numerical representations and indexes, and stored for your search and chat needs. These are permanently destroyed immediately when you delete them. They may live in secure caches for upto thirty days after deletion, but these caches are also soon destroyed.

We are a custodian of your data. Chats contain user ideas, thoughts, and confidential client information. We recognize this and embrace the great responsibility that comes with it. jhana does not use, train AI models with, or share any user chats. Chats are stored in a secure database, and are only accessible to you. They are permanently destroyed immediately when you delete them. jhana admins only access feedback data voluntarily shared by users such as % of chats liked or disliked and feedback messages you may submit to help us improve the platform.

jhana employs a comprehensive, multi-layered data security focused approach. We employ state-of-the-art security features including IAM authentication, verification mechanisms against identity theft, strong database-level encryption, role-based access controls, and SSL/TLS encryption over a secure HTTPS connection. We continuously update our security measures in an ever-evolving digital landscape. We are also prepared for hostile attacks with mitigation measures including robust middleware architecture to vigilantly gatekeep data flow, reverse proxies to counter DDoS attacks, and our content delivery network (CDN) distributes data across geographically dispersed servers.

jhana is built by world-class quantitative academics with expertise in tying AI to ground truth. We only publish models that we would feel comfortable using for our own legal work and carefully test all our models. But AI is a nascent and speedily improving science, and surely, we will stumble. We commit to being transparent about issues, and improving when our models fail to be accurate. Our AI will not always get it right, but we design its interface to make its claims easily verifiable and backcheckable.

jhana's AI is designed to be transparent and easily verifiable. Any claims by our AI include an in-line source link that takes you to the page of the original document that is being quoted. Our outputs also typically include quotations and Standard Indian Legal Citations (SILC) that explain the AI's reasoning. If you encounter an error or inaccuracy, please let us know by clicking on the feedback button below the output or the bug report button in the top right corner of the screen.

No. jhana is a legal research and writing tool for experts and professionals, not a legal advice tool. jhana is designed such that those proficient with the law can easily verify outputs for their fidelity and usefulness, which laypersons cannot. None of the generations of our tools constitute legal advice. While we recognize that some non-legal users might find jhana helpful to understand their lawyer's advice better, jhana should always be used in conjunction with a lawyer's advice, and never as a substitute for it.