| Mobile Web

Concerns grow over side effects of delegating authority to AI agents, operators urged to build controls

Concerns are rising that AI agents acting on users’ behalf could cause harm when users cannot verify or control delegated authority. Experts at a user protection conference in Seoul urged operators to let users set and revoke permissions, require prior approval for high-risk actions such as payments, and keep records and rollback options. Speakers also said liability should not be shifted solely to users, and called for clearer responsibility in linked services and more effective, verifiable safeguards.