A report tells that 4 companies can control AI agents

The Open Markets Institute, a policy group, made a new report on the risk from AI agents. Agents are software that makes selections and does tasks for a user. The report tells that Google, Amazon, Microsoft and Meta control the cloud, models, operating systems, browsers and search engines that agents use. The report gives no market numbers, only an argument.

How it works

The report names 3 causes that can give the market to a small number of companies:

• Own systems: the companies push agents on their operating systems, browsers and search engines.
• Data lock-in: agents keep user data and records, and users find it not easy to go to other agents.
• Control of access: the companies that control standards select which vendors can use the platforms.

The evidence

It points to 2 examples in Europe. Regulators gave agents access to Android from Google and WhatsApp from Meta.

What is not known

The authors want 5 steps, for example data portability, a duty of loyalty to the user and antitrust law steps. European competition authorities work on access to AI tools. US lawmakers start to look at a law with a duty for agents. The report has no information from the 4 companies.

Insight: The agent control software can become a new gate

Gartner, an analyst company, names Microsoft a leader in agent management platforms. These tools control agents from many vendors. Gartner tells that they will control 60% of agents by 2029. Microsoft is also 1 of the 4 in the report. This reads as a new gate: the company that controls the control software can set conditions for other agents. Find out if the platform of each vendor controls agents from other vendors.