Antitrust enforcers across major jurisdictions are turning their attention to so-called stealth mergers and acqui-hires in artificial intelligence — deals structured to escape merger review while still concentrating critical technology and talent.

The development, reported by JD Supra, signals a coordinated shift in enforcement priorities: regulators no longer treat the acquisition of an AI team, without the formal purchase of a company, as outside their reach.

The concern is straightforward. Large technology firms can absorb an AI startup's engineers, founders and intellectual property through licensing deals or hiring sweeps without triggering the notification thresholds that normally force a merger filing. The result, enforcers argue, is consolidation of the AI sector by stealth.

What counts as a 'stealth merger'?

The term covers transactions deliberately designed to avoid scrutiny. Typical structures include:

• Acqui-hires, where a dominant firm hires most of a startup's team rather than buying the company outright;
• Founding-team licenses, where a firm licenses technology from a startup while its founders join the acquirer;
• Partial acquisitions, where a stake is taken below control thresholds but still confers decisive influence over strategy.

Each structure delivers the commercial substance of a merger — talent, technology and competitive capacity changing hands — without the formality of one.

Why is AI the flashpoint?

Artificial intelligence concentrates value in people and models rather than in factories or installed customer bases. A handful of researchers can carry the core of a competitive threat to an incumbent. Acquiring that team can neutralise a future rival before it ever reaches the market.

That dynamic makes acqui-hires more consequential in AI than in most industries. A startup that loses its founding engineers may retain a legal existence but lose its competitive significance. Enforcers on multiple continents now argue this is precisely the kind of transaction merger control was built to catch.

What does the enforcement shift change?

The practical change is jurisdictional reach. Authorities are increasingly prepared to examine transactions that fall below traditional turnover-based notification thresholds, using:

• Call-in powers, allowing agencies to review non-notifiable deals on their own initiative;
• Transaction-value thresholds, which capture high-value acquisitions of firms with little revenue;
• ** strengthened theories of harm**, focused on the elimination of nascent competitors and innovation competition.

For dealmakers, the message is that structuring an AI transaction to avoid filing no longer guarantees avoidance of review. Regulators can — and increasingly will — reach back to examine what was actually acquired.

What should acquirers do now?

Companies pursuing AI talent or technology deals should assume visibility. Practical steps include:

• Assessing whether an acqui-hire or license-plus-hire structure delivers the substance of an acquisition;
• Mapping which jurisdictions' call-in powers could apply, even absent a filing obligation;
• Documenting the competitive rationale, since agencies will test whether the deal removes a nascent rival;
• Building delay into deal timelines for possible informal or formal regulatory inquiries.

The strategic logic

The enforcement posture reflects a broader consensus that AI markets risk tipping early. If foundational models and the talent to build them consolidate into a handful of incumbents within the technology's first commercial decade, later divestiture remedies will be ineffective. Prevention, enforcers have concluded, is cheaper than cure.

For AI startups, the shift cuts both ways. It may preserve exit options beyond absorption by a giant. It may also slow the talent deals that fund many founders' next ventures.

For incumbents, the era of quiet consolidation in AI is closing. Any transaction that moves a meaningful share of a startup's team, technology or direction now carries antitrust risk — whether or not it is called a merger.