🔍 Read the full analysis: How Should We Think About AI Agents Acting 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 should answer when AI agents make decisions on a person’s or organization’s behalf. It separates the question of who pays for a mistake from who must explain and repair it, and who bears moral responsibility. The book argues that existing liability rules and contracts may leave important gaps, particularly for financial losses.
Thorsten Meyer has released Many Hands: Who Answers When the Agent Acts in Your Name?, an ebook examining how responsibility is divided when AI agents buy, book, pay or accept terms for people and organizations. Published in October 2026, the book argues that legal liability, organizational responsibility and moral blame can fall on different parties—an issue that grows as software is given authority to take actions, not just provide answers.
Meyer introduces the issue with a composite scenario: a purchasing agent accepts a steel supplier’s terms, securing a seven percent lower price but committing a company to 120 tonnes over a year, with no cancellation right. The commitment is worth about €370,000. The company owner had already agreed to divide the next year’s volume with another supplier, but the agent was not given that information. Meyer says the agent followed its rules and recorded its actions, even though the result conflicted with the owner’s plan. The scenario is illustrative, not a reported real transaction.
The book organizes accountability around three questions: who pays, who must explain and fix what happened, and who should have acted differently. Meyer says the first is shaped by law and contracts, the second by an organization’s internal processes, and the third is a moral question. Those answers need not point to the same person. The book’s practical proposals include assigning a named person to each agent, limiting permissions to the task, recording decision-making, giving affected people a way to raise concerns, and using records to improve systems rather than simply punish staff.
Meyer also examines the limits of product-liability rules and the effect of vendor contracts. He writes that European product-liability rules cover software but do not cover pure financial losses such as a mistaken payment or a bad contract. He also says some model-provider terms limit liability to roughly a year of fees. These are claims made in the book; the supplied material does not provide the relevant statutory text, contract examples or legal analysis needed to independently assess their reach.
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
When Agent Errors Become Business Losses
The question matters because an agent can commit an organization to a purchase, payment or set of terms before a person has reviewed the result. In the steel example, a lower unit price does not by itself make the decision sound: the volume commitment conflicts with an existing plan, and the company may have no right to cancel. The example illustrates how an apparently successful task can create a larger operational problem.
When something goes wrong, the person closest to the outcome may be the easiest to identify, even if decisions about the system’s design, permissions or oversight were made elsewhere. Meyer’s concern is that responsibility can be displaced onto an employee or operator who had limited ability to detect or prevent the error. His proposed safeguards—clear authority limits, named oversight and usable records—are aimed at making it easier to identify who made a decision and to respond to its consequences.
The book’s distinction also helps readers avoid treating “accountability” as one question with one answer. A company may have a contractual route to recover money, an internal duty to investigate, and a separate ethical obligation to apologize or change its practices. Those remedies can diverge, leaving affected people uncertain about who will act even when the agent’s decision is documented.
Products Worth Considering
Dell UltraSharp 32 U3226QER 4K Thunderbolt Hub Monitor
Apple Mac mini Desktop Computer with M5 Pro Chip (2026)
Logitech MX Keys S Wireless Keyboard, Graphite
As an affiliate, we earn on qualifying purchases.
From Answering Questions to Taking Action
Meyer says agents now do more than generate responses: they can book, buy, pay, accept terms and change data in someone else’s name. He connects the problem to Dennis F. Thompson’s 1980 formulation of the “problem of many hands,” in which numerous contributors make it difficult to assign responsibility for an outcome. In Meyer’s account, agents add distance and speed to that existing problem.
The book uses earlier cases to show how responsibility can become concentrated on a nearby human while broader system choices remain less visible. It discusses the 2018 fatal crash involving an Uber self-driving test vehicle in Tempe, Arizona. Meyer cites the U.S. National Transportation Safety Board’s account that the system detected pedestrian Elaine Herzberg 5.6 seconds before impact, did not settle on what she was, and, by design, withheld emergency braking for one second. The investigation identified several contributing factors; the criminal case resulted in the vehicle operator pleading guilty to endangerment. Meyer connects the pattern to anthropologist Madeleine Clare Elish’s term “moral crumple zone,” for a human who absorbs blame around a technological system.
Another chapter covers the Dutch childcare benefits scandal, in which families were wrongly required to repay benefits. Meyer says the prime minister described responsibility as joint after the cabinet resigned in January 2021, and cites a September 2026 government report recognizing more than 44,000 affected people. These cases are not evidence that AI agents caused those events; they provide context for Meyer’s argument that complex systems can diffuse responsibility.
“protects the integrity of the technological system, at the expense of the nearest human operator.”
— Madeleine Clare Elish, as quoted by Meyer
The Reach of Liability Rules
The supplied material does not include the full legal analysis behind Meyer’s discussion of European product liability, so the exact boundaries of the rules and how they apply to particular agent decisions remain unclear. A contract dispute, financial loss and physical injury may be treated differently, and outcomes depend on the facts and applicable law. The statement that some providers cap liability at about a year of fees is also presented without named contracts or a survey of terms.
The book’s opening steel purchase is explicitly a composite scenario, not a verified case involving a named company. The material also does not establish how often agents currently make comparable commitments, how frequently those decisions cause losses, or whether the proposed governance measures have been tested at scale. Meyer’s claims about accountability identify risks and questions; they do not establish that every agent deployment has the same legal or organizational weaknesses.
Practical Questions for Agent Deployments
The immediate development is the ebook’s release; the supplied announcement gives no schedule for further editions, events or policy proposals. For organizations using agents, the questions raised by Meyer’s framework are practical: what can the agent commit to, who set those limits, who can stop an action, and who reviews a decision after it affects someone else?
Meyer’s suggested controls point to records that identify a responsible person, permissions scaled to the job, decision logs and a route for affected people to challenge an outcome. Whether these measures are sufficient will depend on how they are implemented and on the applicable law and contracts. The broader policy and legal questions—particularly how responsibility should be assigned when an agent causes a financial loss—remain unresolved in the material provided.
Key Questions
What is Many Hands about?
Thorsten Meyer’s October 2026 ebook examines who is responsible when an AI agent acts in someone’s name. It separates legal responsibility for payment from organizational duties to explain and fix a problem, and from moral responsibility for a decision.
Is the steel-purchasing example a real incident?
No. Meyer describes Theresa and the purchase as a composite scenario built from common situations. The source material does not identify a real company or transaction.
Does the book say AI agents are legally responsible for their actions?
The book asks who should answer when an agent acts for a person or organization. Its framework focuses on the people, organizations and contracts around the system; the supplied material does not claim that an AI agent itself has legal responsibility.
What safeguards does Meyer propose?
He proposes naming a responsible person for each agent, limiting its permissions to the task, keeping records of decisions, providing a way for affected people to raise concerns, and using logs to improve systems.
What remains unsettled?
The exact legal treatment of agent-caused financial losses, the reach of specific provider contracts, and how well proposed safeguards work in practice remain unclear from the supplied material. It also does not establish how often agent mistakes cause losses.
Source: ThorstenMeyerAI.com
