Telescope Logo
Telescope Logo

AI agents and consumer liability: new interfaces, new risks

Read time: 8 minutes


As AI assistants and autonomous agents become an interface through which consumers engage, AI will reshape relationships between consumers and businesses.

In the context of this article, consumers are individuals contracting with businesses on a personal basis and not for business purposes.

Across the globe, a guiding consumer law principle is that consumer law and compliance do not change simply because AI is used.

People walking being tracked

As AI assistants and autonomous agents become an interface through which consumers engage, AI will reshape relationships between consumers and businesses.

In the context of this article, consumers are individuals contracting with businesses on a personal basis and not for business purposes.

Across the globe, a guiding consumer law principle is that consumer law and compliance do not change simply because AI is used.

Old rules, new interface

The UK’s Competition and Markets Authority (“CMA”) guidance of March 9, 2026, on complying with consumer law when using AI agents states that consumer law does not change because AI is used. The guidance applies wherever AI agents carry out consumer-facing functions, such as customer service, sales, marketing and refunds. The CMA’s key message is that AI agents are treated in law in the same way as human staff.

The existing framework therefore continues to apply in the UK. Harder questions arise in practice. If an agent summarizes key terms, has the consumer received the required information clearly before being bound? If a consumer delegates a purchase, what amounts to informed consent? When interaction is mainly through a third-party assistant, who appears to be the “trader”? These questions remain unsettled, creating risk.

Where the new risks lie

The risks become most visible as AI agents move from making suggestions to taking action:

  • Misleading recommendations: an agent that misstates price, features or availability may amount to a misleading action or omission. The CMA expects agents to be trained and configured so that they do not behave in a misleading way.
  • Automated purchasing: transactions made without clear authorization, or a renewal the consumer did not anticipate, invite complaints, chargebacks and regulatory attention.
  • Personalization: tailored offers that exploit a consumer’s vulnerability or create artificial urgency may cross the line into unfair practices.
  • Accountability gaps: A business that deploys an AI agent to deal with consumers on its behalf remains responsible for that AI agent’s conduct, even where the AI system comes from a third party or the underlying model was developed externally. The CMA states that words using AI do not reduce legal accountability.

A business that deploys an AI agent to deal with consumers on its behalf remains responsible for that AI agent’s conduct, even where the AI system comes from a third party

The CMA has identified agentic AI as a regulatory priority. Its guidance sits alongside its new enhanced enforcement powers under the Digital Markets, Competition and Consumers Act 2024, exposing businesses to direct enforcement, significant penalties (of up to 10% of global turnover) and reputational damage. The CMA’s focus is accountability, transparency, fairness and trust, rather than restricting innovation. Further changes are expected with the new UK subscription rules coming into force in January 2027, and will affect providers of recurring consumer services.

The CMA has identified agentic AI as a regulatory priority. Its guidance sits alongside its new enhanced enforcement powers under the Digital Markets, Competition and Consumers Act 2024, exposing businesses to direct enforcement, significant penalties (of up to 10% of global turnover) and reputational damage. The CMA’s focus is accountability, transparency, fairness and trust, rather than restricting innovation. Further changes are expected with the new UK subscription rules coming into force in January 2027, and will affect providers of recurring consumer services.

The US has no federal equivalent of the CMA guidance, and consumer protection is divided between federal and state regulation. However, that does not mean there is no consumer protection framework that applies to AI systems. Section 5 of the Federal Trade Commission Act (“FTC Act”) prohibits unfair or deceptive acts or practices and applies regardless of whether a human or an AI agent makes the statement or takes the action. The states also have consumer protection statutes with similar scope, and some go further than the FTC Act. For example, the State of Utah explicitly states that it is not a defense to a consumer protection violation that generative AI made the statement or undertook the act. The US approach is more fragmented than in the UK, and federal and state efforts are increasingly in tension. At the federal level, the emphasis is on applying existing consumer protection law rather than adopting new AI-specific rules. For example, the FTC has proposed a policy statement indicating that failing to disclose personalized pricing based on consumers’ personal data is likely to be unfair or deceptive under Section 5 of the FTC Act. States, by contrast, are adopting AI-specific rules. Colorado, for example, has replaced its original AI Act with a narrower law, effective January 1, 2027, which requires developers and deployers of automated decision-making technology used in consequential decisions to give consumers advance notice and post-decision disclosures, and grants consumers certain rights. The federal government is seeking to limit these state efforts, including through the December 2025 executive order and a call for federal pre-emption legislation; although state consumer protection laws of general application are expected to continue to apply. Businesses serving US consumers should therefore track state requirements alongside general federal and state unfair and deceptive practices law.

Singapore

In May 2026, the Infocomm Media Development Authority (IMDA) published a discussion paper on legal responsibility for AI agents. The paper examines how existing civil liability and private law frameworks under Singapore law, notably contract and negligence, may apply when autonomous AI agents cause harm.

The paper identifies the different actors involved in developing and deploying AI agents, including model developers, tooling and platform providers, system providers, deployers and end users. A majority of the working group considered that existing legal principles may address many cases, but identified practical challenges in establishing intention, foreseeability and causation, particularly where multiple agents or parties are involved. Determining how liability should be allocated among these parties may also be difficult.

Separately, the Monetary Authority of Singapore (MAS) issued its Guidelines on Artificial Intelligence Risk Management on October 7, 2026. The Guidelines emphasize that financial institutions remain accountable for AI used in the services they provide, including AI developed, operated or supplied by third parties.

Hong Kong

In August 2026, the Office of the Privacy Commissioner for Personal Data (PCPD) published guidance on protecting personal data privacy in the use of agentic AI. The guidance clarifies that AI agents are not legal persons and that organizations, as data users, remain responsible for complying with the Personal Data Privacy Ordinance (PDPO), including its six Data Protection Principles, when AI agents collect, hold, process or use personal data.

Although the guidance is non-binding, it reflects the PCPD's expectations for compliance with the PDPO. Beyond privacy, Hong Kong has no AI-specific consumer liability regime. Claims arising from the use of AI agents would therefore generally be assessed under existing laws, including contract, tort (such as negligence) and the unfair trade practice provisions of the Trade Descriptions Ordinance.

China

On September 7, 2026, the Supreme People’s Court issued its first judicial guidance dedicated to AI disputes, the Opinions on Lawfully Adjudicating Cases Involving Artificial Intelligence Disputes. Rather than creating new statutory rights or obligations, the Opinions guide courts in applying existing laws, including the Civil Code, the Personal Information Protection Law and the Consumer Rights Protection Law, to AI-related disputes.

The Opinions clarify that fault-based liability generally applies to AI-related torts unless the law provides otherwise. It also indicates that product liability principles are directed primarily at AI products embodied in physical goods. In addition, the Opinions address issues including AI-generated content that infringes personality rights or disseminates false statements and the use of personal information in AI training.

The Opinions further address algorithmic price discrimination. Businesses that use algorithms to impose unreasonable differences in prices or other transaction terms for the same goods or services may incur liability where such practices infringe the lawful rights and interests of consumers and cause harm.

Australia

Australia continues to rely primarily on existing and technology-neutral laws. The Treasury’s October 2025 review concluded that the Australian Consumer Law (ACL), combined with other laws, can generally handle the challenges presented by AI-enabled goods and services.

Liability for AI agents is therefore likely to be assessed under the ACL (including the prohibitions on misleading or deceptive conduct and the consumer guarantees), product liability rules, privacy law and common law negligence. The ACL prohibits businesses from contracting out of applicable consumer guarantees. Accordingly, a business cannot exclude liability for breaching those guarantees simply because it used an AI system to act on its behalf.

Korea

Korea's AI Basic Act and its Enforcement Decree took effect on January 22, 2026. The Act requires providers of generative and high-impact AI products and services to notify users of the use of those technologies. It also imposes transparency requirements for AI-generated content, including clear labeling where the content is difficult to distinguish from reality.

However, the Act primarily establishes a regulatory framework for AI rather than a dedicated consumer compensation regime. Consumer claims arising from AI agent failures are therefore likely to continue to be addressed under existing civil law, consumer protection, product liability and sector-specific regimes. The Act may also influence the standard of care expected of AI developers and deployers.

Across the region, the common principle is that deployers remain responsible for compliance and for mitigating the risks of the AI agents they put in front of consumers.

Implications for design and deployment

We recommend that compliance is built into each consumer facing digital item or service. The CMA is clear that agents are not “set and forget” and businesses need ongoing oversight and rapid responses, especially where harm can scale. Businesses should not overstate an AI’s capabilities or role, and customers should understand how it is used. Customer journeys should include clear purchase confirmation, disclosures that remain intact when passed on by an agent, and terms reflecting how the service works.

Mobile phone icons

Key takeaways

Businesses contracting with consumers can capture the benefits of agentic technology without taking on unmanaged risk. Steps to take now include:

  1. Map your agents. Identify every consumer-facing touchpoint where AI speaks or acts for you, including those supplied by third parties.
  2. Test and monitor. AI agents should be trained and configured to comply with consumer law requirements and businesses should ensure that robust testing and ongoing monitoring is in place once AI agents are deployed. Businesses need human oversight and ensure that processes are in place to correct errors or unintended behavior.
  3. Revisit contracts. Make sure supplier and platform agreements allocate responsibility and audit rights in a way that reflects your continuing liability to consumers.
  4. Be transparent. Tell customers when and how AI is involved, and do not overstate what it can do.

Ultimately, the businesses best placed to benefit from agentic AI will be those that treat consumer protection not as a constraint on innovation but as a core part of responsible design and deployment.

Key takeaways

Businesses contracting with consumers can capture the benefits of agentic technology without taking on unmanaged risk. Steps to take now include:

  1. Map your agents. Identify every consumer-facing touchpoint where AI speaks or acts for you, including those supplied by third parties.
  2. Test and monitor. AI agents should be trained and configured to comply with consumer law requirements and businesses should ensure that robust testing and ongoing monitoring is in place once AI agents are deployed. Businesses need human oversight and ensure that processes are in place to correct errors or unintended behavior.
  3. Revisit contracts. Make sure supplier and platform agreements allocate responsibility and audit rights in a way that reflects your continuing liability to consumers.
  4. Be transparent. Tell customers when and how AI is involved, and do not overstate what it can do.

Ultimately, the businesses best placed to benefit from agentic AI will be those that treat consumer protection not as a constraint on innovation but as a core part of responsible design and deployment.

Authors


Eve England

Eve England

Partner, UK

View profile
Matthew Gough

Matthew Gough

Partner, UK

View profile

Alby Yuen

Partner, Head of Technology & Digital, Asia

View profile
Hannah Winiarski

Hannah Winiarski

Special Counsel, US

View profile

Steffen Morawietz

Principal Associate, Germany

View profile

In this edition


In this edition


Children on phones

Age assurance as digital infrastructure

Read time: 7 minutes

Read more

Illegal content in the AI supply chain

Read time: 6 minutes

Read more

The content licensing reset

Read time: 5 minutes

Read more

Platform governance across multiple rulebooks

Read time: 5 minutes

Read more

Creators, advertising, and monetization

Read time: 6 minutes

Read more

Protecting premium content at internet speed

Read time: 6 minutes

Read more

Disputes and enforcement in the digital content sector

Read time: 4 minutes

Read more

Previous page

Next page

Arrow

© Eversheds Sutherland. All rights reserved. Eversheds Sutherland is a global provider of legal and other services operating through various separate and distinct legal entities. Eversheds Sutherland is the name and brand under which the members of Eversheds Sutherland Limited (Eversheds Sutherland (International) LLP and Eversheds Sutherland (US) LLP) and their respective controlled, managed and affiliated firms and the members of Eversheds Sutherland (Europe) Limited (each an "Eversheds Sutherland Entity" and together the "Eversheds Sutherland Entities") provide legal or other services to clients around the world. Eversheds Sutherland Entities are constituted and regulated in accordance with relevant local regulatory and legal requirements and operate in accordance with their locally registered names. The use of the name Eversheds Sutherland, is for description purposes only and does not imply that the Eversheds Sutherland Entities are in a partnership or are part of a global LLP. The responsibility for the provision of services to the client is defined in the terms of engagement between the instructed firm and the client.

Connect with us

linkedin logo
facebook icon
youtube icon