AI-Assisted Legal Self-Representation: Epistemic Autonomy, Access to Justice, and the Limits of Procedural Delegation

Legal rights may be formally available yet difficult to exercise without reliable legal knowledge. AI-assisted self-representation raises a question of procedural authority as well as epistemic autonomy: when may technology help a person act, and when would its use change who is authorized to act? A tier-sensitive account distinguishes assistance with acts individuals may lawfully perform themselves, legislative enlargement of those acts, and delegation of reserved acts to an artificial representative. The distinction supports a conditional allocation of justificatory burdens. Where the citizen remains the authorized actor, restrictions imposed specifically because assistance is computational require reasons tied to the risks of that assistance, while existing rules on professional services, privacy and procedure remain applicable. Extending personal-action rights requires evidence that legal protection will not deteriorate; conferring authority on an artificial representative additionally requires a valid source of entitlement and a reason to prefer that status over feasible accountable alternatives. Concerns about legal-aid inequality, human control, accountability, legal agency and interoperability bear differently on each institutional boundary. It does not establish an existing general right to use AI in legal proceedings, validated safety, cost savings or the novelty of its constituent principles. Voluntary assistance is justified only where users retain meaningful control, sources can be checked, human support remains accessible and remedies are effective.