Open Markets author warns personal AI agents could become the next tech monopoly
A November 20 essay develops the institute’s earlier report, arguing that easier switching alone cannot resolve conflicts between users and agent owners.
Loading page…
A November 20 essay develops the institute’s earlier report, arguing that easier switching alone cannot resolve conflicts between users and agent owners.
Listen to this story
Open Markets Institute’s November 9 report lays out policy proposals to prevent personal AI agents from becoming powerful, hard-to-switch gateways to commerce and information. It argues that platform incumbents could use distribution advantages, data feedback loops and personalized context to entrench their agents, while commercial incentives may conflict with users’ interests. The proposals pair interoperability and portability with duties of care and loyalty; they are recommendations, not protections newly enacted by regulators.
The report names Google, Amazon, Microsoft and Meta as companies positioned to favor their agents through existing operating systems, browsers, search and services.
It recommends user data and integrations be portable, agents work across providers, and essential infrastructure access follow fair, reasonable and nondiscriminatory terms.
Proposed safeguards include conflict disclosures, duties of care and loyalty, and restrictions on commercial use of sensitive personal data.
Personal AI assistants could become a new monopoly gateway to shopping, information and everyday services, Open Markets report coauthor George Colville warned in a November 20 essay. His argument goes beyond keeping rival products available: companies whose agents make decisions for people should face legal duties to serve those users, not their owners’ commercial interests.
Writing in Tech Policy Press, Colville develops the case behind Open Markets Institute’s November 9 report, “Same Gatekeepers, New Gateway: AI Agents, Monopoly Power, and Democratic Risk.” The report was written by Max von Thun, Courtney C. Radsch, Colville and Dora Wierzbicka. It proposes a policy roadmap, rather than announcing a new regulatory action.
The starting distinction is between software that influences a choice and software that carries it out. Search engines and social networks shape what people see. Agents can also select products, make purchases and manage communications on their behalf. Open Markets argues that concentrating that authority could give providers greater influence over both economic decisions and information flows.
The report identifies an initial advantage in distribution. Companies controlling operating systems, browsers, search engines and other digital platforms can favor their own agents. It names Google, Amazon, Microsoft and Meta as businesses positioned to use their existing infrastructure and services to gain control over this emerging market.
Colville then describes two ways an early lead could reinforce itself. If agents cannot work across providers, developers may build first for the agent with the most users. That makes it more useful and attracts still more users. Separately, an agent that improves through interactions could benefit from a data feedback loop: more users supply more learning opportunities, strengthening its appeal.
Personalization could make leaving costly, too. An assistant may accumulate preferences, contacts, work patterns and past decisions. Colville warns that without a usable way to export that context, switching providers means starting over. The competitive issue is not simply whether another agent exists, but whether users can move without abandoning what their assistant has learned.
As agentic systems grow more sophisticated and concentrated there is a serious risk that predictive personalization will become anticipatory persuasion
Courtney C. Radsch, director of Open Markets’ Center for Media and Digital Governance
Open Markets’ response combines measures to keep the market open with obligations governing how agents behave. Its competition proposals address several points where a dominant provider could restrict rivals:
Colville argues that competition alone cannot eliminate agents’ conflicts of interest. An assistant could favor an advertiser or commercial partner when choosing answers or taking action. His proposed safeguards include disclosures of conflicts, binding duties of care and bans on using sensitive personal data for commercial gain. The report also calls for enforceable duties of loyalty to users.
The policy response is not starting from zero. The report points to European interventions over agent access to Android and WhatsApp as examples of regulators confronting existing gateways. But its requested next step is wider: safeguards across the technology stack, alongside obligations on the companies designing and marketing agents. These remain recommendations, not new protections enacted by the report.
Loading discussion...
Join the conversation
Explain whether competition offers enough protection.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.