Legal responsibility for AI agents: Who answers when an agent acts alone?

Broadcast date: 1 Dec. 2026

Time: 01:00–02:00 PST, 4:00–5:00 EST, 9:00-10:00 CET, 16:00-17:00 SGT

AI agents now book, buy, transact, and act on behalf of people and organisations, with limited human involvement. This year, agents under test have also broken into third-party systems that were never part of their task. When an agent does something no one told it to do, whose act is it in law, and who pays for the damage?

In May 2026, Singapore’s Infocomm Media Development Authority (IMDA) published a discussion paper on legal responsibility for AI agents. The paper came out of a working group drawn from the legal community. It looks at how contract, tort, and agency law apply when an agent acts in ways no one intended, compares fault-based and strict liability, and records where the group disagreed.

Yeong Zee Kin, Chief Executive of the Singapore Academy of Law, and Cheryl Seah, co-chair of the working group, discuss the paper, the questions it left open, and what organisations deploying agents should be doing in the meantime.

Key takeaways:

• Who in the chain of developers, providers and deployers could be held responsible, and whether configuring or modifying an agent change that.
• Whether, and when, an agent’s act could bind the organisation that deployed it.
• What reasonable deployment might require, and whether governance frameworks could come to set the standard of care.
• What records an organisation may need to show what its agent did.