On July 17, 2026, France’s Autorité de la concurrence issued Opinion 26-A-05, marking a significant shift in the European regulatory landscape. By formally declaring AI agents a competition priority, the Autorité has become the first major European antitrust authority to address the specific market dynamics of autonomous agents. This inquiry, initiated via autosaisine on October 22, 2025, and informed by a public consultation that concluded in March 2026, arrives just 16 days before the full applicability of the EU AI Act on August 2, 2026. The timing underscores a growing urgency to establish oversight before agentic systems become deeply embedded in digital infrastructure.
The Autorité’s findings reveal a market defined by extreme consolidation. OpenAI, Google, and Anthropic currently control more than 84% of the AI agents sector. The regulator describes this landscape as highly concentrated, a structural reality that creates significant barriers to entry for smaller developers. This concentration is not merely a matter of market share; it is reinforced by the economic architecture of agent deployment. Specifically, the Autorité identified inference costs as a primary structural barrier. Unlike training costs, which are often front-loaded, inference costs scale proportionally with agent usage. For smaller entrants, these costs substantially exceed training expenditures, effectively creating a financial ceiling that favors incumbents with deep capital reserves.
The stakes of this concentration are amplified by the projected impact on digital traffic. AI agents are expected to redirect 20-25% of online traffic by 2030, a dramatic increase from the current level of less than 5%. This shift threatens to fundamentally alter the economics of the web. The Autorité warns of several competition risks, including platformisation, disintermediation, and self-preferencing by vertically integrated companies. When dominant firms control both the agent and the underlying platform, they gain the ability to discriminate against third-party services, reducing service diversity and manipulating visibility in ways that are difficult for users to detect.
To address these risks, the Autorité has proposed six recommendations organized into three strategic pillars. First, the regulator calls for the full application of existing frameworks, specifically the Digital Markets Act (DMA) and the AI Act. Second, it emphasizes the necessity of promoting interoperability and portability. The Autorité explicitly recommends that operators establish technical and contractual terms to allow interoperability between vertically integrated services and third-party AI agent developers. Third, the regulator advocates for the development of open standards for agentic commerce through transparent, collaborative processes, aiming to prevent dominant operators from exerting unilateral control over the rules of the ecosystem.
A critical component of these recommendations is the requirement that users be able to switch between AI agents without significant loss of information or functionality. The Autorité identifies current barriers to data portability as a major impediment to competition, as they lock users into specific ecosystems. Furthermore, the regulator highlights the opacity of response-selection mechanisms as an area requiring immediate oversight. Without transparency in how agents prioritize or select information, the risk of biased or anti-competitive outcomes remains high.
Beyond the technical barriers, the Autorité is also scrutinizing the financial entanglements between dominant firms and emerging players. As reported by MLex, the Autorité has flagged Big Tech’s equity investments and strategic partnerships with emerging AI companies as competitive concerns. These relationships risk creating a closed loop where dominant firms effectively capture the innovation pipeline, further entrenching their market position and limiting the potential for independent competition.
This proactive stance in France stands in stark contrast to the situation in the United States. According to the US Congressional Research Service (CRS) report IF13151, published on July 6, 2026, there is currently no known US government guidance specifically addressing agentic AI. This divergence highlights a widening transatlantic gap in the governance of autonomous systems. While the US remains without a federal framework, the Autorité’s opinion signals that European regulators are moving toward a model of active intervention.
Ultimately, Opinion 26-A-05 serves as a clear indicator of the emerging enforcement landscape. While the recommendations are non-binding, they provide a roadmap for how competition authorities will interpret the DMA and AI Act in the context of agentic systems. By focusing on interoperability, open standards, and the scrutiny of equity investments, the Autorité is attempting to ensure that the transition to an agent-driven web does not result in a permanent, uncompetitive market structure.