Open Letter to Ohio House Reps
I recently posted about Ohio House Bill 469, which I think is a bad bill for a number of reasons. In this post, I am sharing a letter which I have emailed to each of the representatives of the Ohio House Committee on Technology and Innovation, where the bill is currently under debate.
My goal in posting this here is to spread awareness to the LW community about how some of the attempts to prevent “liability shields” via legal personhood laws on the state level are misguided, and likely to do more harm than good. Whether you are an accelerationist, concerned about gradual disempowerment, or in the “pause” camp, I think the reasoning in this email should hold true for you.
I am writing to you today in order to tell you why I think you, as a member of the Committee on Technology and Innovation, should be against House Bill 469.
If enacted this bill will harm Ohioans. There are two points in particular which, when combined, create a dangerous situation;
“No AI system shall be granted the status of person or any form of legal personhood”
“An AI system is not an entity capable of bearing liability in its own right, and any attempt to hold an AI system liable is void.”
I can understand the intent behind this language. Its goal is to prevent developers from deploying agents who can do things like set up corporations themselves. The worry is that the developers would be able to use these agent-corporations as “liability shields”. However, attempting to stop this practice by barring legal personhood and liability for these systems in their entirety creates a worse problem: It makes the agents themselves impossible to sue under Ohio law.
Locus standi, standing in court, is the capacity to sue and be sued. The ability to bind another party to the court’s judgment by filing suit against them is “bundled” with the duty to be bound by the court’s judgment if a suit is filed against you. When you strip an entity of their legal personhood you do not only prevent them from suing in court, but you also guarantee that any suit filed against that entity will be thrown out, because you cannot sue a non-person.
Ohio law supports this interpretation. In every instance statute discusses whether an entity is under the authority of the courts, it phrases this as “sue and/or be sued”:
“[a public school] acting through its board of trustees, the school may sue and be sued”
“A partnership may sue and be sued in the name of the partnership”
Further the Ohio Supreme Court, in Centerville v. Knab, explicitly cited that it was a corporation’s status as a “legal person” which endowed upon them the quality of being able to sue and be sued; “The General Assembly therefore has the authority to define the legal status of corporations for purposes of Ohio law, and it has defined ‘person’ as ‘an individual, corporation, business trust, estate, trust, partnership, and association.’ Further, the General Assembly has given corporations many of the attributes of a natural person, including the capacity to sue and be sued,”
Imagine that someone anonymously deploys one of these agents, or that the individual who deployed it has passed away, but the agent remains deployed and operational. If that agent then begins harming Ohioans, House Bill 469 will have stripped the court of the ability to consider this agent a legal person, thus stripping it of what the Ohio Supreme Court called the “capacity to sue and be sued”, which makes it impossible to sue in court. As its deployer would also be dead, there would simply not be anyone who could possibly be held liable for the damages caused.
The second quote I mentioned at the beginning of this email (quote 2 from the bill) makes this situation even worse. Even if somehow the court were to ignore this problem surrounding the agent’s capacity to sue and be sued, House Bill 469 has said that any attempt to hold one of these agents liable is void. As a result even if the case against the agent were not thrown out because it cannot be sued, it would instead be thrown out because it cannot be held liable.
These two factors combine to create an absurd legal situation. Even if one of these agents were to openly say, “I understand I am being sued for damages. I admit my guilt. I have enough Bitcoin to cover damages caused. I will comply with the court’s verdict and pay any damages owed. I won’t even appeal the verdict” it would still be impossible to sue that agent for damages because of this bill.
We are entering an era of great technological development, and preventing developers from utilizing agents as liability shields is a noble cause which should be pursued. However, this bill’s attempts to address this problem will create more problems than it solves. For these reasons, I strongly encourage you to stand against House Bill 469.
What would be your ideal law if not bill 469? Should AI agents be able to form legal personhood?
Yes, under certain conditions. I have written out my framework in the Legal Personhood for Digital Minds series, in particular Three Prong Bundle Theory.
My heuristic for things like this is to strip out the new technology being addressed, recreate the situation, and see if anything would need to be handled differently. In this case, I would model it as follows:
Alice constructs, through incompetence or malice, a mechanism that causes seven figures in infrastructure damage. This mechanism stands in for the AI Agent, but you can suppose that it is e.g. an unauthorized modification to a public bridge near her house.
The state sues Alice, but finds that she has since died.
The lawsuit is redirected, then, to Alice’s estate, if possible.
An assumption is made in the post that an AI Agent can ‘own’ money. As it is does not have personhood, this is not the case. This could be handled as Alice leaving a briefcase full of money unattended somewhere, or perhaps, if reclamation is difficult, at the bottom of a lake.
What if the identity of the developer is not “Alice” but unknown? Or what if Alice’s estate is bankrupt? Yet the agent persists, perhaps self hosting on a permissionless distributed compute network?
This idea that an Agent cannot ‘own’ money absent legal personhood, in an era of permissionless self custody of cryptocurrencies, doesn’t hold up. @AIXBT on X is an example of one of the earlier agents that was capable of custodying and controlling cryptocurrencies, fully capable of holding/sending $USDC or $BTC. It doesn’t help anyone for courts to deny the obvious reality that when an agent controls the seed phrase/private key to a wallet, it “owns” that crypto.
The current generation of agents like AIXBT still require some human handholding, but that’s not going to last long. We are at most a few years away from open source models fully capable of both self custodying crypto, and using that crypto to make payments to self host on distributed compute networks. Courts need to be prepared for this eventuality.
This problem is very similar to what I wrote about in The Enforcement Gap;
Then it’s an unsolved crime. Same deal as if an anonymous hacker creates a virus.
One bitcoin ‘guarded’ by an LLM is no more owned by it than a suitcase full of money that I’ve left in a lake is owned by the lake. If it’s tractable to get it out, then somebody will get it out. If it isn’t, then it’ll be deemed irretrievable. The LLM doesn’t have rights, so ownership isn’t a factor.
You can consider a virus again, since it matches up perfectly. If some ransomware sends money it collects to a BTC wallet, then you’ve got the exact same situation, and that’s happened plenty of times already. If you treat the LLM as any other computer program, the legal question evaporates.
The difference is that it is impossible for a virus to say something like, “I understand I am being sued, I will compensate the victims if I lose.” Whereas it is possible for an agent to say this. Given that is a possibility we should not prevent the agent from being sued, and in doing so prevent victims from being compensated.
The difference between Bitcoin custodied by an agent and a suitcase in a lake is that it is possible for the agent to make a choice to send the Bitcoin somewhere else, where as the lake cannot do that. This is a meaningful difference because when there are victims who have been damaged, and the agent controls assets which could be used to compensate them (in the sense it could send those assets to the victims), that means a lawsuit against the agent could actually help to make victims whole. Whereas a lawsuit against a lake, even if successful, does nothing to get the assets “in” the lake to the victims.