Data and confidentiality
Where client data goes, privilege, and locally run models.
Where does our client data actually go? Does it leave the firm?
It depends on the work, and zedders is clear about exactly where it lands. Public chatbots can keep and reuse what you put in, so they are off the table for client material. The safer paths are AI under enterprise contract, or models that run on your own infrastructure so data never leaves the firm. For privileged material, keeping it inside the firm is the default.
Is it even safe to put privileged client information into AI?
Into a public tool like consumer ChatGPT, no. It can expose confidential material and put privilege at risk, and the law society's guidance is explicit against it. The safe paths are AI with proper contractual protection and data isolation, or models that run on infrastructure the firm controls. zedders designs confidentiality in from the start rather than bolting it on later.
What are locally-run models, and do we actually need them?
A locally-run model runs on infrastructure your firm controls, so sensitive data doesn't need to leave the building to be useful. You do not always need one: enterprise plans for hosted platforms are available, though expensive, and for work not covered by client confidentiality, like checklists and research, well-governed cloud AI is sufficient. For privileged material, local models remain the strongest answer.
How do you keep our use of AI compliant and defensible?
By building the obligations in rather than treating them as an afterthought: collecting only the data a task needs, handling it onshore where required, and keeping a person in the loop to verify anything AI produces, which is what the courts and conduct rules require. What zedders builds is designed to fit your firm's own AI policy, not work around it. If you do not have one yet, that is something zedders can help you shape.
What happens to our data if we stop working with you, who owns it?
You own your data and the systems we build for you. If the relationship ends, any client data we hold is returned or deleted on request and documentation is handed over so you can run things independently, with no reliance on us.
What does zedders do with our firm's data?
To build something that fits, zedders needs to see how your firm works, and that can include client matter data. That data is kept confidential, used only to build and improve your systems, and never for another firm or purpose. Where your data trains those systems, it stays on models kept inside the firm or otherwise contained, so confidentiality holds. A confidentiality agreement governs all of it, and client data is returned or deleted when the work ends.
Send it over and we will answer it directly, usually the same day.