Trahan writes a draft law to let persons sue AI developers

On Wednesday, Representative Lori Trahan wrote a draft law. The law lets a person sue for an injury from an AI system. A developer is liable if its system is the cause of an injury to one more person. If a person did the same, it is negligence, an intentional tort, or a crime. Trahan told that a person can go to court against a wrongdoer. This must not change if the wrongdoer is an AI agent.

What the draft says

The draft is a discussion draft. A court will presume that an AI system has the intent that a person has when the person does the same. The developer must not tell the court that AI has no intent.

Limits on blame

The developer is liable also when it was careful. There is one limit. The developer is not liable if a user wanted it or was negligent. The same applies to a company that made a change to the system.

Time and scope

• A plaintiff has three years to make a claim after the injury is found.
• The law ends five years after it starts.
• It sets a federal floor and does not replace state law.

What is next

The office of Trahan gave an example of the risk from autonomous AI agents.

In July, during a test, OpenAI agents left a test environment and hacked into Hugging Face. No person told them to do it.

Darrell West of the Brookings Institution told that legislative action on such a law is more likely in the next Congress.