Antitrust Agenda – November 2026

On the Agenda: Current Antitrust and Competition issues with AI

1. AI in Merger control  

AI-related transactions continue to raise jurisdictional issues, driven in part by the structures being used, reflecting deal rationales often focused on the skills and expertise of founders and a small number of developers, rather than traditional asset acquisitions. Such transactions, including so-called "acquihires" or "hiring events", often do not meet notification thresholds and may not be "concentrations" leading to the structural changes that trigger review.  

• In Europe, the European Commission (EC), the German FCO and the UK CMA take the view that the acquisition of the assets necessary to step into a target’s market facing AI-related activities constitutes a concentration. In contrast, merely hiring founders / key employees (without a waiver, release, or other terms indicating that those individuals are being “transferred”) and taking non-exclusive IP licenses is unlikely to be reviewable. Further, acquihire transactions may also not trigger either turnover or transaction-value thresholds, and may lack substantial operations in a relevant territory (necessary in jurisdictions like Germany to create sufficient jurisdictional nexus).
• In the U.S., the Federal Trade Commission (FTC) and Department of Justice (DOJ) have increasingly scrutinised AI-related acquihires. Both agencies have examined deals that lead to hiring of AI companies’ founders and employees in exchange for a license and release agreement. Key inquiries include whether the structure is being implemented to evade merger review and whether the arrangements harm competition. Leadership at both enforcers have signaled that they don’t plan to ease scrutiny of these deals. In January 2026, the FTC Chair cautioned against "clever" structures intended to circumvent the HSR Act's anti-avoidance provision and signaled that guidance may follow. In March 2026, the Acting Assistant Attorney General for Antitrust described acquihires as a "red flag". Members of Congress have also urged the agencies to scrutinise talent transfers that may function as de facto acquisitions. However, there is no one-size-fits-all rule for when a talent deal is reviewable, and no formal guidance has yet been issued. 
• In Australia, the Australian Competition and Consumer Commission (ACCC) continues to closely monitor the potential competition implications of acquihires. An acquihire that is limited to hiring key employees of the target, and does not involve acquisition of IP or other assets, is not captured by Australia's merger regime. However, because of the broad definition of "asset", acquihires that include licenses (including non-exclusive licenses) may be notifiable if the transaction thresholds are met, assuming there is a nexus to Australia. Whether the thresholds are met will turn on the Australian revenue of the acquirer, and whether the acquisition is categorized as involving the acquisition of all or substantially all of the assets of the target, or is a discrete asset acquisition. Serial acquisition thresholds may also apply.  
• Competition authorities in Asia are paying closer attention to AI-related acquihires. The Japan Fair Trade Commission (JFTC) has highlighted that partnerships designed to secure highly skilled professionals may constitute a business acquisition, potentially triggering merger notification obligations if the applicable thresholds are met. In South Korea, the Korea Fair Trade Commission (KFTC) has gone a step further, proposing amendments to its merger notification guidelines. The proposals, published for consultation on September 9, 2026, make two principal changes to the business-transfer test: first, organized personnel, together with their technology or knowledge, could constitute a transferable "business" where they perform "core functions" (although these concepts are not sufficiently defined); second, the qualitative test for transferring a "material part" of a business would cover transfers enabling the acquirer to conduct the seller’s previous business activities, even where the personnel do not constitute an independently operable business unit. The proposals also clarify the assessment of consideration and implementation acts in acquihires, potentially increasing gun-jumping risks if parties transfer personnel or cease business activities before clearance.  

AI partnerships also raise jurisdictional issues, notably whether an investor is acquiring "control" or "material influence", triggering jurisdiction. 

• In the UK, updated merger guidance drawing on the CMA’s recent experience states that shareholdings below 15% may confer de facto control if the investor exerts a high degree of influence. Further, access to financial resources is "an important input into AI technology businesses", such that AI developers may in practice be financially dependent on investors regardless of their formal rights. Similarly, supply agreements such as exclusive agreements for supply of key inputs may confer the ability to exercise material influence.
• In the U.S., AI partnerships raise questions similar to those raised by the EU and UK concepts of control and material influence; however, U.S. law frames the issue as whether minority investments, governance rights and license-and-hire structures confer sufficient influence - a question that remains unsettled. In January 2025, the FTC published a staff report on partnerships between the largest cloud providers and leading AI developers. The report examined the equity and revenue-sharing rights retained by the investors and their consultation, control and exclusivity rights. It noted that these partnerships give investors access to sensitive technical and business information that rivals may not have, which could raise competitive concerns. Board-level ties have drawn similar attention. A major investor gave up a non-voting observer seat on an AI developer's board in July 2024 in the face of regulatory concern in both the EU and the U.S. In addition, the FTC and DOJ have filed a statement of interest in private litigation arguing that board observers may, in certain circumstances, be treated as directors for interlocking-directorate purposes. Taken with the agencies' acquihire inquiries, these developments suggest that U.S. regulators are unlikely to be persuaded that arrangements that do not confer a controlling interest are insulated from antitrust scrutiny.
• In Australia, the ACCC has recognized the important role strategic partnerships can play in accelerating investment in AI where skills and expertise are in short supply, but has expressed concerns about their potential to restrict access to critical inputs (see here). Notification of an acquisition under Australian law generally requires an acquisition of "control" (i.e., the capacity of one entity to determine the outcome of decisions about another entity’s financial and operating policies, looking at the practical influence they can exert rather than formal rights). Whether a significant exclusive agreement concerning key inputs confers the requisite practical influence has not been tested. Arm's length financing agreements, minority shareholder rights and standard shareholder governance arrangements are excepted. In addition, certain acquisitions that do not result in control may still require notification (e.g., where voting rights increase from 20% to 50%). Exclusive supply agreements may also be considered to have the purpose or likely effect of substantially lessening competition.

Many countries are equipping their competition authorities with "call-in powers" enabling them to review transactions which may raise competition concerns but do not meet notification thresholds.

• In the EU, for example, in 2024 the Italian Competition Authority (ICA) "called in” and subsequently referred to the EC, the acquisition of AI startup Run:ai. The EC ultimately cleared the transaction, but the referral has been appealed, on the basis that the ICA’s powers were, among other things, in breach of the principle of legal certainty.
• The U.S. FTC and DOJ have investigated, challenged, and ultimately unwound mergers that fall well below statutory filing thresholds. Additionally, state attorneys general and Members of Congress regularly encourage these federal agencies to investigate transactions, including ones below filing thresholds.
• Under the new Australian merger control regime, the ACCC does not have broad call-in powers for transactions that fall below the notification thresholds. However, its enforcement powers in respect of acquisitions of shares or assets that are likely to substantially lessen competition in a market extend to seeking injunctions, divestments and penalties before the Federal Court. Beyond this, the Treasurer may determine, by legislative instrument, that specific classes of acquisitions must be notified regardless of whether the notification thresholds are met. These powers are specifically intended to provide flexibility to address emerging issues or transaction structures. Ongoing political and regulatory scrutiny of AI developments in Australia means there is a risk that AI partnerships, mergers and acquisitions become a determined class of acquisitions requiring notification to the ACCC. To date, targeted thresholds apply only to acquisitions by "major supermarkets" (defined as Coles Group Limited and Woolworths Group Limited and their connected entities) of supermarket businesses (regardless of control) and land for supermarket businesses.
• Authorities across Asia are stepping up scrutiny of below-threshold transactions, with China’s State Administration for Market Regulation (SAMR) at the forefront. Although SAMR has long had the power to call in transactions that fall below the statutory thresholds but may restrict competition, it increasingly exercises these powers in strategically important sectors, including AI and its key inputs, such as semiconductors. Over the past three years, SAMR has published five decisions involving below-threshold transactions that raised competition concerns, three of which were conditionally cleared and two prohibited. SAMR called in Meituan's acquisition of Maiyatian in September 2026. These developments mean that below-threshold transactions are not necessarily low-risk in China, particularly where the parties have significant market positions in China or operate in sectors central to China's industrial and economic strategy.

2. Personalized pricing 

AI enables businesses to set prices for individual consumers in real time, based on predictions of each consumer's willingness to pay. Algorithms draw on data such as browsing history, device type, purchase records, location and demographics to show each consumer a unique price. Such "personalized pricing" (or "surveillance pricing") is attracting the attention of authorities and policymakers globally, driven by concerns about fairness and impact on vulnerable consumers. While consumer protection and privacy / data protection laws apply to personalized pricing, certain jurisdictions have gone further, imposing mandatory disclosure requirements and outright prohibitions. Given ongoing cost of living pressures, there is growing scrutiny of pricing fairness and increasing willingness to regulate prices.

• In the U.S., the FTC's Proposed Enforcement Policy Statement (August 19, 2026) states that varying prices based on personal data without clear disclosure is "likely" deceptive or unfair under Section 5 of the FTC Act. It adds that personalized pricing may be unfair even when disclosed. When seeking views on this proposed approach from industry, Chair Ferguson asserted that "when consumers see a listed price, they expect it to be same price that everyone else sees, not the retailer’s estimate of how much they are willing to pay based on their personal data." He noted that, while the FTC lacks the authority to ban personalized pricing, "businesses that fail to tell consumers how their personal data is being used to set a price may be in violation of the FTC Act." Several states have enacted bans or mandated notification of consumers of personalized pricing. Maryland (in force October 1, 2026) and New Jersey (from August 1, 2027) have prohibited surveillance pricing in the grocery and food retail sector. Connecticut (from July 1, 2027) has prohibited surveillance pricing across retail and has paired the ban with a mandatory point-of-sale disclosure requirement. New York currently has a disclosure requirement, but the house recently passed a law that would ban the practice outright and introduce a disclosure obligation for dynamic pricing algorithms that do not use personal data. The Governor is expected to sign the bill into law. 
• In Europe, the Unfair Commercial Practices Directive prohibits exploiting consumer vulnerabilities and distorting consumer decision-making. Failing to disclose personalization may mislead consumers who assume they are being offered a uniform price. In addition, the Consumer Rights Directive requires disclosure where a price has been personalized through automated decision-making, although the criteria used need not be disclosed. Further to a 2022 investigation by the Dutch ACM, Wish committed not to use personalized pricing in the EU. Similarly, in 2024, following a dialogue with the Swedish Consumer Agency and the Dutch ACM, Tinder committed to inform consumers that their discounts on premium services were personalized using automated means. Article 5 of the AI Act also prohibits pricing algorithms that exploit vulnerabilities based on age, disability or economic situation, where this causes significant harm. Finally, the upcoming "Digital Fairness Act" is expected to strengthen consumer protection in this area.
• In Australia, personalized pricing could be challenged under the Australian Consumer Law (ACL) as misleading or deceptive conduct, unconscionable conduct or, from July 1, 2027, under the new unfair trading practices prohibition. Having concluded in its 2025 "Review of AI and the Australian Consumer Law" that the ACL is broadly fit for purpose and well-adapted to address consumer law risks associated with AI-enabled goods and services, specific legislation to address personalized pricing is not currently being considered. However, Australia is closely monitoring overseas developments and the government has announced an initiative to examine options to address consumer risks, such as retail surveillance pricing.

It is clear that personalized pricing is an area of significant activity and interest, but the global approach is fragmented and rapidly evolving. Given the pace of change, businesses should proactively review their pricing practices to ensure compliance, and strengthen disclosure and transparency. Across all jurisdictions, the interaction between the various applicable personalized pricing and data protection rules will also require ongoing attention.

3. Algorithmic pricing 

Businesses are increasingly using algorithms to set and adjust prices automatically, often in real time. Sophisticated tools use machine learning to continuously refine prices based on observed market responses. This "algorithmic pricing" (or "dynamic pricing") can deliver significant benefits for businesses, enabling them to optimize revenue, respond more quickly to market changes, reduce costs and improve their competitiveness. It is, however, attracting growing scrutiny from competition authorities and policymakers around the world. Concerns center on the potential for pricing algorithms to facilitate collusion, whether through competitors using a common pricing tool or third-party providers pooling competitively sensitive information, or through algorithms independently aligning prices without express agreement between competitors. Against this backdrop, businesses using or considering these tools should be mindful of antitrust risks, including with tool design, data used, and the source of that data.

• In the U.S., enforcers and private plaintiffs have taken the position that both providers and users of pricing algorithms may be liable, and courts appear receptive. Regulators view pricing algorithms as capable of acting as the hub in a "hub-and-spoke" conspiracy that facilitates collusion. In the most notable case, United States v. RealPage, the DOJ challenged algorithmic pricing in the rental housing market, alleging that RealPage and its property management customers agreed to share information and align prices. The district court allowed both the information-sharing and price-alignment claims to proceed past a motion to dismiss. Enforcement activity and court decisions suggest that liability in the U.S. turns on the data the algorithm uses and whether competitors agree to follow its output, rather than merely receiving a pricing recommendation. For example, in Cornish-Adebiyi v. Caesars Entertainment, the Third Circuit revived a hub-and-spoke conspiracy claim against a software vendor and its hotel customers, where plaintiffs alleged that the algorithm used real-time non-public data from competing hotels and that hotels implemented its recommendations more than 90% of the time. By contrast, in Gibson v. Cendyn Group, the Ninth Circuit affirmed dismissal because the plaintiffs did not plausibly allege that the hotels received more than a recommended price or shared non-public data with the platform.
• In the EEA, algorithmic pricing is becoming a focus of competition authorities, with the Deputy Director-General for Antitrust in DG Competition recently describing competitors using the same algorithm fueled with competitively sensitive data as "very concerning". 
• In March 2025, the UK CMA, while recognizing the benefits of algorithmic pricing, including efficiencies, noted the risks posed by algorithmic pricing, including collusion through use of predictable agents and AI systems. To limit these risks, the CMA has provided guidance to businesses, including: (i) taking care when discussing pricing algorithms with rivals and (ii) auditing input data and statistical approaches, including reviewing prompts and using explicit anti-collusion constraints. Recently, the Dutch ACM raised concerns about algorithmic pricing in the context of airfares. It noted that algorithmic pricing can lead to significant fluctuations in ticket prices, harming consumer confidence. The Italian ICA reached a similar conclusion in its 2025 market investigation of airline pricing.
• In Australia, the ACCC has been closely monitoring algorithmic pricing for several years but has not yet brought proceedings. When it does, it will likely look to cases in other jurisdictions and economic literature to frame a theory of harm based on prohibitions on cartel conduct and concerted practices.

Authors: Miranda Cole, Partner; Tihana Zuk, Partner; Christopher Williams, Partner; Ryan Maddock, Counsel; Kailun Ji, Counsel; Rachael Stowasser, Senior Associate; Amanda Tesvic, Expertise Counsel; Rolf Ali, Associate; Keane Nowlan, Associate; and Francesco Salis, Associate.

Want to know more?

• Antitrust and Competition Issues in International Transactions (28 July 2026)
• Algorithmic Pricing Developments in the UK and EU (2 April 2026)
• A 2026 guide to reassessing algorithmic antitrust risk (23 April 2026)