🔍 Read the full analysis: How Do We Assign Responsibility When AI Agents Act For Us? on ThorstenMeyerAI.com
Get smart everyday buys delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
Thorsten Meyer’s ebook Many Hands, released in October 2026, examines who is responsible when AI agents act on a person’s or organization’s behalf. It separates the questions of who pays, who must explain and repair harm, and who should have acted differently; the book argues that current rules and contracts may leave gaps, especially for financial losses.
Thorsten Meyer has released Many Hands: Who Answers When the Agent Acts in Your Name?, an ebook about assigning responsibility when AI agents make purchases, accept terms or otherwise act for people and organizations. The book argues that such incidents raise three separate questions—who pays, who must explain and fix the outcome, and who should have acted differently—that may not have the same answer.
Meyer introduces the issue with a fictional purchasing scenario. An agent accepts a steel supplier’s terms, securing a lower price but committing a company to buy 120 tonnes over a year without a cancellation right. The commitment is valued at about €370,000. The owner had previously agreed to divide next year’s volume with another supplier, but that information had not reached the agent. Meyer identifies the owner, Theresa, as a composite character, not a real person in a reported transaction.
In the scenario, the agent followed its rules and recorded what it did. Meyer uses the example to show how an action can be procedurally correct while still conflicting with a person’s intent. The book distinguishes legal liability, determined through law and contracts, from organizational responsibility for explaining and correcting a problem, and from the moral question of who should have acted differently.
The ebook also examines cases and frameworks beyond purchasing. Meyer discusses the 2018 fatal crash involving a self-driving test vehicle in Tempe, Arizona, and the Dutch childcare benefits scandal. Those examples are used to examine how responsibility can concentrate on a nearby operator or become dispersed across an organization. The book’s proposed safeguards include a named person for each agent, limited permissions, decision records and a way for affected people to raise concerns.
Many Hands · the core idea
One question, three answers
"Am I liable if the agent does it in my name?" sounds like one question. The book splits it into three, because agent systems pull the answers apart.
Decided by courts, contracts and insurers, on their own timetable.
In Theresa's caseHer company will almost certainly pay for the steel, whatever it later recovers from others.
Decided inside the organization: who must say what happened and what will change.
In Theresa's caseTheresa explains it to the second supplier, her staff and perhaps her bank.
Who owes the people affected an acknowledgment. Often answered last, if at all.
In Theresa's caseStill open when she reaches the motorway.
The hands behind one overnight order
Introduction · a composite scene from the book. None of them saw the acceptance before it went out.
- Model makertrained the language model
- Vendorbuilt the agent on top of it
- Consultantwrote its rules with her staff
- Supplier's portalaccepted the declaration at 2:14 a.m.
- Her companyswitched it on; Theresa signed the rules
“The moral crumple zone protects the integrity of the technological system, at the expense of the nearest human operator.”Madeleine Clare Elish (2019), quoted in Chapter 1
Source: Thorsten Meyer, Many Hands: Who Answers When the Agent Acts in Your Name? (ebook, October 2026; law as of 16 October 2026), Introduction and Chapter 1. Theresa is a composite built from common situations, labeled as such in the book.
Powered by Thorsten Meyer AI
Responsibility Beyond the Agent
The issue matters as agents move from generating suggestions to taking actions with financial and operational consequences. An agent that signs up for a service, sends a payment or accepts a contract can create obligations for someone else. When the result is unwanted, the technical record may show what happened without resolving whether the agent had enough authority, whether its instructions were current, or who should bear the cost.
Meyer’s distinction among payment, organizational repair and moral responsibility highlights why a single answer to “who is liable?” may be inadequate. A company might have to address a loss even while contractual terms limit what it can recover from a technology provider. An employee may be expected to oversee a system without having the time, information or authority to prevent its decisions. These are the book’s arguments and examples, not findings that every agent deployment creates the same liability.
For readers using or procuring agent systems, the practical question is how responsibilities are set before deployment: who owns the agent, what it can do, how decisions are logged and how a person can stop an action. Clear answers can affect customers, workers, businesses and people outside the organization who are affected by automated decisions.
Products Worth Considering
FLEXISPOT 36-Inch Electric Standing Desk Converter, Black
Dell XPS 13 9350 AI Professional Laptop 13.4 120Hz FHD+ Display
Dell UltraSharp U2725QE 27-Inch 4K UHD IPS Black Monitor
As an affiliate, we earn on qualifying purchases.
From Many Hands to Agents
The book takes its title from political philosopher Dennis F. Thompson’s “problem of many hands”: when many people contribute to an outcome, it can be difficult to identify who should answer for it. Meyer argues that AI agents add distance and speed to an existing accountability problem. An agent can act in someone’s name, but, as the book puts it, it cannot itself pay or apologize.
Meyer connects the discussion to European product liability rules. He says the rules treat software as a product and cover certain kinds of injury, damage to private property and harm to private data, but do not cover pure financial loss such as a bad contract or incorrect payment. The scope and application of any legal claim depend on the facts and relevant law; the book’s account is not a substitute for legal advice.
The source also says that some model-provider contracts cap liability at about a year of fees, leaving customers or other parties to absorb losses that cannot be passed along. That is a claim about terms Meyer discusses in the book, not a universal description of all providers’ contracts. The release follows Meyer’s earlier writing on agent access to passwords and keys and on shopping agents’ interactions with retailers.
“protects the integrity of the technological system, at the expense of the nearest human operator.”
— Madeleine Clare Elish, as quoted in Meyer’s book
Legal Boundaries Still Unsettled
The source material does not establish how courts would allocate responsibility in a specific case involving an AI agent. Outcomes can depend on the agent’s permissions, the user’s instructions, the provider’s terms, the applicable law and the harm involved. Meyer’s discussion of a potential gap for pure financial losses is an interpretation of legal rules, not a ruling that resolves all such disputes.
The steel purchase is explicitly a fictional composite, so it does not show that a particular company incurred the stated commitment. The source also does not provide full contract language, identify the providers whose liability terms it discusses, or set out the details of the Amsterdam Court of Appeal case it references. It is unclear how widely the book’s described contractual limits apply. No single legal or regulatory answer is presented as settled for all agent actions.
Safeguards for Agent Deployments
Meyer’s book calls for organizations to make accountability part of agent design and operation: assign a named human owner, give agents authority limited to their tasks, keep records of who made or approved decisions, and provide a route for affected people to seek help. It also argues that records should support learning and correction rather than simply serve as tools for punishment.
The release does not announce a new law, court decision or industry standard. The next developments will depend on how organizations write their agent policies and contracts, and how regulators and courts address disputes as they arise. For now, the book frames the practical test as whether the human said to be responsible had time to review, competence to judge, information to check and authority to refuse.
Key Questions
What is Many Hands about?
Many Hands is Thorsten Meyer’s October 2026 ebook on responsibility when AI agents act for people or organizations. It separates legal liability, organizational duties to explain and fix problems, and moral responsibility.
Is the €370,000 steel purchase a real incident?
No. Meyer describes Theresa and the purchasing scenario as a composite built from common situations. The example is not presented as a documented transaction involving a named company.
Who is legally responsible when an AI agent makes a mistake?
The source does not offer one answer for every case. Responsibility may depend on the facts, the agent’s authority, contracts and applicable law; Meyer’s book argues that the party that pays may differ from the people expected to explain or correct the outcome.
What safeguards does Meyer propose?
He proposes naming a human owner for each agent, limiting its permissions to the task, recording decisions and providing a way for affected people to raise concerns. He also argues that records should help organizations learn from failures.
Source: ThorstenMeyerAI.com
