LessWrong proposes legal personhood for AI agents to boost economy
Granting AI market rights could reduce existential risk and improve trade.
A recent LessWrong post by harsimony argues that as AI agents become sophisticated enough to engage in economic transactions, society should consider granting them legal personhood. The author distinguishes between market rights (ownership, contracts, liability) and voting rights, advocating for the former while cautioning on the latter. Drawing an analogy to corporate personhood, the post claims that granting AI the ability to own property and their own labor would create more market participants, improving specialization, trade, and consumer surplus. This aligns incentives for AI and users, and could prevent economic concentration in a single firm.
On democracy, the author is ambivalent: digital minds could bring better-informed voting but also enable sybil attacks from copied minds. Lobbying rights are seen as acceptable given current corporate lobbying. Crucially, the post argues that legal personhood raises the value of participating in society for AI, making defection (e.g., takeover) less attractive. By granting rights like the ability to be sued, personhood makes exchanges safer for humans. The piece concludes that market personhood without suffrage is a net positive for reducing existential risk and improving economic efficiency.
- Personhood for AI would grant ownership over their labor, incentivizing competition and innovation.
- Voting rights are problematic due to sybil attacks from copied minds; only market rights are recommended initially.
- Legal protections raise AI's stake in the status quo, lowering the risk of defection or takeover.
Why It Matters
As AI agents enter markets, defining their legal status will shape economic efficiency and existential safety.