Senior Counsel: Emerging trends in competition law and AI – LexisNexis
LexisNexis’s May senior counsel session, hosted in collaboration with Cripps LLP and Flex Legal, explored emerging trends in competition law and their practical implications. The session focused on three areas attracting increasing regulatory attention: competition law in labour markets, competition risks arising from minority stakes, and the developing relationship between competition law and agentic AI.
The discussion opened with a reminder of the core purpose of competition law: to protect the competitive process, allowing markets to operate effectively and deliver better outcomes for consumers. For businesses, the risks of getting this wrong remain significant. Competition law breaches can lead to fines of up to 10% of the business’ turnover, compensation claims where individuals have suffered, exclusion from public contracts, potential criminal consequences for individuals, and serious brand damage. As the speaker noted, competition law may be a ‘narrow’ risk area in the sense that businesses may not encounter it every day, but where a breach occurs, the consequences can be severe. The CMA’s own guidance similarly highlights both corporate and individual consequences for unlawful anti-competitive conduct.
Labour markets: competition law is not just about customers
A central theme of the session was the growing focus on competition law in labour markets. Historically, there has been a perception that competition law primarily concerns agreements between businesses about customers, prices, markets or suppliers. However, this is not the case, as competition law also applies to agreements between employers about wages or working conditions.
This is particularly important because businesses do not need to compete in the same product or service market to be competitors for competition law purposes in a labour market context. Two companies in entirely different sectors may still be competing for the same categories of employee, contractor or freelancer. The CMA’s 2025 guidance, Competing for talent, expressly addresses this point, explaining that businesses may compete to hire or retain workers even where they do not compete for customers.
The session identified three key categories of risk in labour markets:
- no-poaching agreements
- wage-fixing
- exchange of competitively sensitive information
These risks apply not only to employees, but also to freelancers, contractors and other categories of worker. The speaker noted that this shift in enforcement focus appears to reflect wider concern about the impact of collusion on workers’ pay, mobility and choice, as well as the broader impact on innovation and productivity where businesses are unable to compete freely for talent.
No-poaching agreements
No-poaching arrangements can take a number of forms. They may involve an agreement not to hire another business’s staff, an agreement not to approach or solicit employees with job opportunities, or an agreement not to do so without the other business’s consent. Importantly, such arrangements do not need to be reciprocal: a one-way commitment can still raise competition law concerns.
The session referenced a widely circulated Steve Jobs email to Adobe as an illustration of the type of conduct that would now be highly problematic. In that email, Jobs complained that Adobe was recruiting from Apple and referred to Apple having a standing policy not to recruit from Adobe. While such exchanges may previously have been viewed by some as commercially robust conduct, the speaker emphasised that similar communications today would trigger serious competition law concerns.
At the same time, the session drew an important distinction between unlawful no-poaching agreements and properly justified non-solicitation clauses in commercial contracts. For example, a consultancy agreement may legitimately prevent a customer from soliciting the consultant’s staff during the term of the engagement and for a limited period afterwards. However, such clauses need to be objectively necessary for the commercial arrangement and proportionate in scope, duration and geographic reach. The CMA guidance makes a similar distinction, recognising that non-solicitation clauses in secondment, consultancy or service-provider arrangements may not breach competition law where they are necessary and proportionate.
For in-house counsel, the practical message is that template non-solicitation wording should not simply be carried forward without review. Clauses should be tested against the commercial rationale for the arrangement, kept time-limited, and drafted no more broadly than necessary.
Wage-fixing and benchmarking
The second labour market risk discussed was wage-fixing. This arises where businesses competing for the same workers agree to fix pay, benefits or other employment terms. It may include agreeing the same wage rate, setting caps on pay increases, or coordinating other terms and conditions.
The speaker also highlighted a more subtle risk: information sharing. Businesses may share pay or benefits information for apparently legitimate reasons, including benchmarking or industry analysis. However, where that information is current, specific, confidential or capable of influencing another business’s pay strategy, the exchange can reduce uncertainty in the market and create competition law exposure.
The session drew a practical distinction between information that is likely to be lower risk and information that is likely to be problematic. Aggregated, anonymised and historic information is generally less concerning, particularly where no individual business can be identified. Publicly available salary data, including data published by the Office for National Statistics or job advertisement platforms, is also less likely to raise concerns. By contrast, direct exchanges between competitors about current or future pay intentions, day rates, benefits or recruitment strategy are high risk. The CMA guidance similarly notes that genuinely public, properly aggregated and historic information is less likely to be competitively sensitive, whereas information that reduces uncertainty or influences strategic decisions may break competition law.
A helpful rule of thumb offered was that if the information would genuinely help a competitor decide what to pay, who to recruit, or how to respond to market pressure, it should be treated with caution.
Regulator crackdown
Regulators have increasingly investigated companies suspected of breaching competition law in the labour markets. As an example, in March 2025, the CMA issued an infringement decision to five sports broadcast and production companies after finding that they had unlawfully shared competitively sensitive information about rates of pay for freelance workers involved in the production and broadcasting of sports content. Four companies agreed to pay fines totalling £4.2 million, while Sky was exempt from a financial penalty because it had alerted the CMA to its involvement before the investigation was launched.
In this particular instance, the CMA found 15 instances where pairs of companies unlawfully shared sensitive information about pay, including day rates and pay rises. The written evidence in that case, including emails and WhatsApp messages, was particularly direct. This serves as a reminder that competition investigations often turn on contemporaneous internal and external communications, and that informal exchanges can be just as damaging as formal agreements.
Other jurisdictions have also increasingly focused on labour-market competition issues. In the US, the Department of Justice announced its first criminal no-poach charges in 2021. In the same vein, in 2024, the European Commission’s 2024 policy brief on labour market infringements confirmed that non-poaching agreements would generally breach competition law.
The speaker also addressed collective bargaining. The CMA recognises that coordination has a legitimate role in collective negotiations between employers and workers or trade unions. However, employers should not use collective bargaining as a reason to exchange competitively sensitive information among themselves unless this is genuinely necessary and cannot be achieved through less risky means, such as independent aggregation and anonymisation. The CMA’s guidance expressly states that it will not seek to enforce competition law against genuine collective bargaining, but that coordination outside that context may become an illegal cartel.
Minority shareholdings: when investment creates competition risk
Another talking point related to competition risks arising from minority shareholdings. The key point was that a minority, non-controlling stake in a competitor does not automatically create a single undertaking for competition law purposes. Unless one company has control over the other, the businesses remain separate actors and must continue to behave independently.
The European Commission’s Delivery Hero/Glovo decision illustrates this point. In this case, Delivery Hero acquired a minority non-controlling stake in Glovo in 2018 and later acquired sole control in 2022. The Commission fined Delivery Hero and Glovo €329 million after finding that, during the period before full control was acquired, the parties engaged in anti-competitive coordination, including a non-poaching agreement, exchange of commercially sensitive information and allocation of geographic markets. The Commission described the case as important because the practices were facilitated through the anti-competitive use of Delivery Hero’s minority stake in Glovo.
This is significant as minority investments are often treated as corporate or strategic matters rather than competition law flashpoints. However, where a minority investment creates board access, information rights, regular management contact or strategic influence, the risk of improper coordination increases.
The practical response is to build competition safeguards into minority investment structures from the outset. This may include carefully limiting information rights, controlling access to competitively sensitive information, and where appropriate, training directors and observers, and ensuring that investment monitoring does not become operational coordination.
Agentic AI: from assistance to autonomous action
Agentic AI is new territory for businesses and regulators alike and is likely to bring up a myriad of legal issues. Further regulatory guidance may follow as the technology develops but some concerns have already been raised by the CMA and other regulators.
Agentic AI differs from the generative AI tools with which many businesses are now familiar. Generative AI systems, such as chatbots, generally respond to prompts and produce content or analysis. Agentic AI goes further: it is given a goal, plans steps to achieve that goal, uses tools, takes action and adapts to changing circumstances.
In March 2026, the CMA published two documents, one being a research paper, Agentic AI and consumers, which explores how agentic AI systems are used and how they are a potential step change in how consumers engage with markets.
These guidelines highlight that the CMA is looking closely at agentic AI from a consumer protection perspective. While the technology may bring benefits, including time savings, improved accessibility and more personalised services, it also raises risks around errors, hallucinations, misleading outputs, manipulation, bias and discrimination.
A particular concern is that businesses may design or deploy AI systems in a way that steers consumers towards outcomes that are more profitable for the business rather than best suited to the consumer. This creates obvious consumer law risk, especially where the consumer believes the AI agent is acting neutrally or in their interests.
The CMA’s guidance, Complying with consumer law when using AI agents, emphasises that the same rules apply whether customers interact with a human or an AI agent. Businesses remain responsible for what their AI agents do, including where the system is designed or provided by a third party. The guidance focuses on transparency, training AI agents to comply with consumer law, monitoring performance, maintaining human oversight and refining systems quickly where problems arise.
Algorithmic pricing and competition law
Although agentic AI is new, algorithmic pricing is not and it is also not inherently unlawful. Algorithms can help businesses respond to demand, reduce costs and improve efficiency. However, regulators are increasingly concerned about circumstances in which pricing algorithms may facilitate collusion or reduce competitive uncertainty.
The session identified several potential risk areas:
- classic collusion, where competitors agree not to undercut each other and use software to implement that agreement
- tacit coordination, where algorithms react predictably to market events and soften competition
- hub-and-spoke structures, where competitors use the same pricing tool, data provider or intermediary
- information sharing, where an algorithm is trained on or influenced by competitively sensitive competitor data
The CMA has previously warned that pricing algorithms can be used to implement or monitor price-fixing, and that rival businesses using the same algorithmic system may indirectly exchange confidential information or coordinate behaviour. The CMA has also made clear that businesses should understand how their pricing tools work and that not knowing what an algorithm is doing is not an excuse.
The speaker gave a practical example of how risk could arise. If several competing businesses use the same third-party pricing tool, and that tool responds to price changes in a predictable way, one business may know that increasing its own prices could lead competitors using the same system to follow. Where enough competitors rely on the same tool, prices may become aligned without any explicit human agreement.
The CMA’s March 2026 blog on AI and collusion reinforces this concern, advising businesses not to share competitively sensitive information directly or indirectly, not to allow competitors’ confidential information to influence pricing, to take particular care when using the same algorithm as competitors, and to scrutinise data and algorithmic approaches where needed.
The CMA has launched its first substantial enforcement action into suspected algorithm-enabled information sharing in the hotel sector, in March 2026. The investigation demonstrates the regulator’s interest in data analytics tools and indirect information-sharing risks. Similar issues are being examined in other jurisdictions. The Dutch and Italian competition authorities have carried out market investigations into airlines which are suspected to have employed algorithmic pricing techniques.
Practical steps
The session closed with a practical reminder that competition law compliance needs to keep pace with changing business practices, technologies and workforce models. Legal teams should ensure:
- competition law risks are understood within the business, including all relevant teams
- training has been provided and controls are in place to protect against the sharing of competitively sensitive information
- there is strong awareness about agentic AI and its risks
- assessments are carried out to determine whether connected companies form a single undertaking for competition law purposes, and controls are in place to manage the sharing of competitively sensitive information between them
The key message was that competition risk increasingly arises outside traditional sales and pricing teams. HR, recruitment, technology, procurement, corporate development and AI governance functions all now sit within the competition law risk landscape.
The big picture
Competition law enforcement is moving into areas that many businesses may not historically have treated as core competition risk. Labour market practices, minority investments and AI-enabled pricing or consumer interactions are now firmly on the regulatory agenda.
The focus is clear: competition compliance needs to be practical, cross-functional and forward-looking. It is no longer enough to focus only on price-fixing, market sharing or customer allocation in the traditional sense. Businesses must also consider how they compete for talent, how they structure strategic investments, and how technology may shape pricing, recommendations and customer outcomes.
