🔍 Read the full analysis: Who Answers For An AI Agent That Acts In Your Name? 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, published in October 2026, examines responsibility when AI agents make purchases, accept terms or otherwise act for people and organizations. It argues that legal liability, organizational responsibility and moral blame can fall on different parties; the source does not report a new law or court ruling resolving those questions.
Thorsten Meyer published an ebook in October 2026 examining who is responsible when an AI agent acts in a person’s or organization’s name, a question with growing practical importance as agents move beyond answering questions to making purchases, accepting terms and changing data. Many Hands: Who Answers When the Agent Acts in Your Name? argues that the person who pays for a mistake, the person tasked with fixing it and the person who deserves blame may not be the same.
Meyer introduces the issue with a fictional composite scenario, not a reported real-world incident. In it, a purchasing agent accepts a supplier’s terms overnight, agreeing to buy 120 tonnes of steel over 12 months for about €370,000. The price per tonne is 7% lower, but the agreement cannot be cancelled. The agent has not been told that the owner had already agreed to divide the next year’s orders with another supplier.
The author says the agent followed its rules and recorded its actions, yet the result conflicted with the owner’s intentions. The scenario is intended to show how an agent can perform its assigned task while acting on incomplete or outdated information. It does not establish that a real company entered this contract or that an AI system caused the described loss.
Meyer frames accountability in three parts: who pays, a question shaped by law and contracts; who explains and repairs the harm, an organizational question; and who should have acted differently, a moral one. The ebook’s examples include the 2018 fatal crash involving a self-driving test car in Tempe, Arizona, and the Dutch childcare benefits scandal. These cases predate the book and are presented as examples of how responsibility can be distributed across people and institutions.
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
Why Agent Decisions Complicate Accountability
The issue matters because AI agents can take actions with consequences beyond an unsatisfactory answer. An accepted contract, a payment or a changed record may create costs or obligations before a person notices what happened. Where decisions pass through software, operators, managers, vendors and business processes, identifying who had the authority and information to prevent a mistake can be difficult.
Meyer’s distinction between legal, organizational and moral responsibility helps clarify why a single answer to “who is liable?” may not settle the practical questions. A court or contract may assign a financial loss to one party, while the organization still needs to investigate and correct the outcome. The person nearest the action may also face blame even when others designed or approved the system.
The book proposes practical controls, including assigning each agent a named human owner, limiting its permissions to the task, keeping records of who made decisions, and providing a way for affected people to seek help. These are the author’s recommendations, not rules shown in the source to have been adopted by regulators or courts.
Popular right now
Vitamix Propel Series 750 Professional-Grade Blender, 4 Programs, Black
Ninja CREAMi 7-in-1 Ice Cream Maker, NC301
Bose QuietComfort Wireless Noise-Cancelling Over-Ear Headphones, Black
As an affiliate, we earn on qualifying purchases.
The Cases Behind Meyer’s Argument
The book draws on the idea of the “problem of many hands,” a term associated with political philosopher Dennis F. Thompson’s 1980 discussion of how shared contributions can make individual responsibility hard to identify. Meyer argues that agents add distance and speed to that problem, while creating an additional participant that cannot itself pay compensation or apologize.
One example is the 2018 crash in Tempe in which a self-driving test vehicle struck and killed Elaine Herzberg. Meyer’s account cites the U.S. National Transportation Safety Board’s findings that the system detected Herzberg 5.6 seconds before impact, did not settle on what she was, and had emergency braking held back by design for one second. The board identified several contributing factors, including the vehicle operator, the company’s risk assessment and oversight, its safety culture, the state and the pedestrian. The source says the operator pleaded guilty to endangerment; Meyer uses the case to examine how responsibility can concentrate on the person closest to an event.
A second example is the Dutch childcare benefits scandal, in which tens of thousands of families were wrongly made to repay benefits. Meyer says responsibility was spread across systems and officials. He cites a September 2026 government report recognizing more than 44,000 affected people, and the Dutch prime minister’s statement after the cabinet resigned in January 2021 that responsibility had to be borne jointly if the whole system had failed. The source also points to European product-liability rules, arguing that their coverage of software-related harm does not extend to pure financial loss such as a bad contract or incorrect payment.
“Protects the integrity of the technological system, at the expense of the nearest human operator.”
— Anthropologist Madeleine Clare Elish, as quoted in Meyer’s book
Liability Depends on the Facts
The source presents an author’s argument and a newly published ebook; it does not describe a new court judgment, government action or industry-wide standard deciding who is liable when an AI agent causes a loss. The legal outcome in any actual dispute would depend on the facts, applicable law, contracts and the parties’ roles. Meyer’s discussion of European product-liability rules is his account of their scope, not a ruling on a particular agent-related claim.
It is also unclear how widely agent providers’ contracts limit liability to roughly a year of fees: the source refers to terms quoted in the book but does not identify the contracts or establish that this cap applies across the industry. The steel purchase is expressly a composite example, so it should not be read as evidence of a specific transaction. The source material also ends partway through its account of a 2023 Amsterdam Court of Appeal case, leaving its details and relevance incomplete.
How Organizations Put Controls in Place
Meyer’s ebook is available as a first-edition ebook, published in October 2026, through retailers listed by the author, including Apple Books, Barnes & Noble, Kobo and Thalia. The accompanying video is described as a five-minute presentation of the book; its narration uses an AI voice clone of Meyer.
For organizations using agents, the immediate questions raised by the book are operational: who authorizes each agent, what actions can it take, what information does it need before acting, and who can stop or review a decision? Meyer recommends naming a responsible person, matching access permissions to the task, preserving decision records and giving affected people a route to raise problems. Whether these safeguards become common practice, and how courts and regulators will treat agent-made commitments, remains unsettled in the source material.
Key Questions
What is Many Hands about?
Many Hands is Thorsten Meyer’s October 2026 ebook about responsibility when an AI agent acts for a person or organization. It separates the questions of who pays, who must fix the outcome and who bears moral blame.
Is the steel contract a real incident?
No. Meyer identifies Theresa and the purchasing scenario as a composite built from common situations. The source does not report that a real company made the described €370,000 commitment.
Does the book establish who is legally liable for an agent’s actions?
No. It discusses how law and contracts may determine who pays, but the source reports no new ruling that settles liability for AI-agent actions. The result in a real dispute would depend on the facts and applicable law.
What safeguards does Meyer propose?
Meyer recommends assigning every agent a named human owner, limiting its permissions to the task, recording who made decisions and giving affected people a way to seek help. These are the author’s proposed practices, not confirmed legal requirements in the source.
Source: ThorstenMeyerAI.com
