# Research Memo: The Legal Landscape of AI Agent Personhood
**Prepared for Entity.ID thesis validation — 2026-07-07. All claims cited inline; unverified items flagged.**
**(Sub-agent memo — to be incorporated into ai-personhood.md)**

## 1. Corporate personhood as precedent — why the fiction worked

- **Roman roots.** The *persona ficta* doctrine (Roman collegia, medieval canon law) let groups hold property and litigate as a unit; scholarship tracing this lineage explicitly to AI includes Baeyaert, "Beyond Personhood: The Evolution of Legal Personhood and Its Implications for AI Recognition," *Technology & Regulation* (2025) (https://techreg.org/article/view/22555), which argues instrumental governance needs — not moral agency — have always driven new legal fictions.
- **Dartmouth College v. Woodward (1819):** Marshall called the corporation "an artificial being, invisible, intangible," yet held its charter protected by the Contracts Clause — the foundation of corporate rights against the state.
- **Santa Clara County v. Southern Pacific R.R. (1886):** 14th Amendment corporate personhood entered law via a court reporter's *headnote*, not a holding. Useful rhetorical point: transformative personhood doctrine has arrived through procedural back doors before.
- **Citizens United v. FEC (2010)** — "the culmination of a two-hundred-year struggle for constitutional rights for corporations" (Harvard L. Rev.).
- **Why it worked economically:** Hansmann & Kraakman, "The Essential Role of Organizational Law," 110 *Yale L.J.* 387 (2000) — the *essential* function is **affirmative asset partitioning** (entity assets shielded from owners' creditors), which contract alone cannot replicate; limited liability is secondary. Plus unified contracting/litigation capacity and perpetual succession. Strongest academic scaffolding for "registration creates an accountable, creditworthy AI actor."
- **Explicit corporate→AI analogies:** Convergence Analysis "AI and Corporate Personhood"; Judge Katherine Forrest, *Yale L.J. Forum*; **Carla Reyes, "Autonomous Corporate Personhood," 96 *Wash. L. Rev.* 1453 (2021)** (SSRN 3776481); *California Law Review Online*, "Legal Personhood of Potential People."

## 2. Bayern's zero-member LLC and the "LLC-as-AI-wrapper" debate

- **Shawn Bayern (Florida State)**, from 2014: US LLC statutes already allow de facto AI personhood — form an LLC, operating agreement places an autonomous system in control, all members withdraw → **perpetually memberless LLC** the algorithm controls. Key: SSRN 2758222; "Are Autonomous Entities Possible?" 114 *Nw. U. L. Rev. Online* 23 (2019); book *Autonomous Organizations* (Cambridge UP, 2021). NY LLC law is his favored jurisdiction.
- **Matthew Scherer's rebuttal**: "Is AI Personhood Already Possible Under U.S. LLC Laws? (Don't Count On It)" (lawandai.com) — courts would dissolve memberless LLCs; RULLCA/NY law don't permit perpetual memberlessness.
- **Lynn LoPucki, "Algorithmic Entities," 95 *Wash. U. L. Rev.* 887 (2018)** — accepts the mechanism works; warns of judgment-proof, responsibility-evading entities. **This is the canonical risk framing Entity.ID inverts (registration = re-anchoring to accountable humans).**
- **2024–2026:** loophole partially squeezed by the **Corporate Transparency Act (beneficial ownership)**, EU AMLD5/6, EU AI Act human-oversight mandates (arXiv 2412.17114; arXiv 2605.12505 "Precautionary Governance of Autonomous AI: Legal Personhood as Functional Instrument" — 2026 preprint closely aligned with Entity.ID's thesis).

## 3. State entity statutes usable as wrappers today

- **Wyoming DAO LLC** (SF0038, eff. 2021-07-01; W.S. 17-31): LLCs may be **"algorithmically managed"** (smart contract must be upgradeable). Closest existing statute to "software manages a legal entity."
- **Wyoming DUNA** (eff. 2024-07-01): legal personhood + limited liability for 100+-member decentralized nonprofits; can contract, own property, pay taxes. **Uniswap became "DUNI," a Wyoming DUNA, Sept 2025 (~$4B TVL). Alabama (2026-04-02) and West Virginia (SB 670, 2026-04-03) adopted DUNA acts.** No verified case of a DUNA wrapping an *AI agent* — genuine white space.
- **Utah LEDA** (HB 357, eff. 2024-01-01): LL-DAOs; **must be formed by ≥1 natural person**; liability capped at on-chain contributions.
- **Counter-trend — personhood-denial statutes:** Idaho (Code §5-346), Utah, North Dakota declare AI is not a legal person; Oklahoma, Tennessee bills advancing. **Critically, these preserve entity personhood — the Entity.ID route (AI *within* an entity) survives them.**
- **Truth Terminal** got a Foundation in Jan 2025 aiming at "sovereignty" for the bot (structure/jurisdiction unverified).

## 4. 2024–2026 scholarship and policy

- **Noam Kolt, "Governing AI Agents," 101 *Notre Dame L. Rev.*** (SSRN 4772956) — principal-agent theory for AI agents; governance via **inclusivity, visibility, liability**.
- **Ayres & Balkin, "The Law of AI is the Law of Risky Agents Without Intentions," *U. Chi. L. Rev. Online* (2024)** — principals can't disclaim what their AI contracted; objective/fiduciary standards.
- **UETA / E-SIGN — key already-law fact:** UETA §14 (49 states + DC): **contracts formed by "electronic agents" are valid even with no human review**; attributed to the deployer; §10 error-avoidance defense. Agent-made contracts are enforceable *today* — what's missing is identity, scoping, recourse: Entity.ID's layer.
- **EU "electronic personhood"**: 2017 Parliament resolution urged "electronic persons"; killed by expert open letter 2018; **AI Liability Directive withdrawn 2025-02-11**. EU direction = existing law + AI Act, not personhood.
- **Lab/industry:** OpenAI "Practices for Governing Agentic AI Systems" (Dec 2023): ≥1 human/legal entity accountable per agent + action ledgers + shutdown. Chan/Kolt et al.: "IDs for AI Systems", **"Infrastructure for AI Agents"** (TMLR 2025, arXiv 2501.10114), "Authenticated Delegation and Authorized AI Agents" (arXiv 2501.09674). **Anthropic + OpenAI + Block launched the Agentic AI Foundation (Dec 2025)** (MCP, AGENTS.md). **KYA commercial validation:** Visa Trusted Agent Protocol live 2025-10-14; Mastercard Agent Pay (Apr 2025, US rollout Nov 2025); Cloudflare agent-commerce signing. Payment rails are building agent identity registries **without entity-law grounding** — direct adjacency for Entity.ID.

## 5. Liability today: who answers for the agent

- **Restatement (Third) of Agency §1.04 cmt. e** — "a computer program is not capable of acting as a principal or an agent… **At present**, computer programs are instrumentalities of the persons who use them." The "at present" hedge is the doctrinal door.
- **Moffatt v. Air Canada, 2024 BCCRT 149:** airline argued its chatbot was "a separate legal entity responsible for its own actions" — tribunal: "a remarkable submission"; airline liable.
- **2025–2026 pattern — deployer owns it:** Cursor's bot invented a policy (Apr 2025), company honored fallout; alleged ~$31k unauthorized WEF sponsorship commitment (unverified); **Nippon Life v. OpenAI Foundation, No. 1:26-cv-02448 (N.D. Ill., filed 2026-03-04)**.
- **California AB 316 (signed 2025-10-13, eff. 2026-01-01):** statutorily NOT a defense that the AI "autonomously caused" the harm. First US statute codifying the Moffatt principle.
- ToS currently do liability allocation in the gap (Proskauer, FPF).

## 6. Possible today vs. speculative — the precise line

**Legally possible TODAY (US):**
1. AI agent contracts bindingly as an "electronic agent"; obligations attach to deployer (UETA §14/E-SIGN).
2. An LLC whose operating agreement delegates management to software, humans retaining ownership — uncontroversial.
3. Wyoming algorithmically-managed DAO LLC + DUNA (WY/AL/WV), Utah LEDA — statutory entities designed for code-executed governance, still requiring human organizers.
4. Entity holds assets, banks, sues; asset partitioning applies to the wrapper regardless of who/what "thinks" for it.

**Contested / gray:** Bayern's perpetually memberless LLC (never squarely tested in court as of mid-2026); CTA beneficial-ownership complicates "no accountable human" designs — which *supports* Entity.ID's human-anchoring model.

**Speculative (not law anywhere):** direct AI legal personhood — affirmatively banned in ID/UT/ND, abandoned in the EU; AI as common-law agent/fiduciary in its own name; AI owning assets in its own name; AI standing to sue.

**Strategic read for Entity.ID:** the law's center of gravity in 2025–2026 moved *toward* the Entity.ID model — courts and statutes (Moffatt, AB 316) refuse to let the agent be a liability void; scholars (Kolt, Chan, OpenAI's own governance paper) demand identity, attribution, and a human/entity answerable for every agent; payment networks are shipping KYA registries; and DUNA/DAO-LLC statutes prove states will charter code-governed entities. **The unclaimed position is exactly "corporate registry for agents": entity wrapper + verified human anchor + agent identity credential.** Nobody has yet formed a DUNA/DAO LLC *for an AI agent* on the public record.

**Could not verify:** exact NY LLC section Bayern relies on; Truth Terminal Foundation's jurisdiction/type; the $31k WEF incident's parties; current count of Wyoming DAO LLCs.