Don't Let AI Become the Internet's Gatekeeper

Jared Sine is the chief strategy & legal officer at GoDaddy.

The internet used to be something we navigated ourselves. Increasingly, AI systems are navigating it for us.

Millions of people already rely on chatbots like ChatGPT and Claude to answer questions, recommend products, summarize news, and help make decisions. Soon, AI agents like Meta’s Muse may compare insurance plans for users, schedule their doctor's appointments, restock their groceries, or rebook a canceled flight before they even open a browser.

For many, the convenience will be undeniable. But so will the power these systems have to steer attention, shape purchasing decisions, and determine which businesses thrive online.

The question is whether this new AI-powered internet will remain open to anyone with a good idea — or whether access to customers and visibility online will depend on the preferences of a few dominant platforms.

We are already beginning to see how this dynamic could evolve. Over the last few years major AI firms have signed licensing agreements with some of the world's leading news organizations. In some cases, these agreements help ensure those outlets appear prominently in AI-generated responses.

These partnerships have the potential to benefit everyone involved. Publishers gain compensation and visibility in an online ecosystem increasingly mediated by AI assistants. AI platforms gain broad access to trusted content and recognizable brands.

In other words, AI companies are beginning to occupy the role search engines once did — deciding which information users encounter first. But AI agents may ultimately wield even greater influence, since they won't just recommend options. They'll act on them.

By 2030, purchases "initiated, influenced, or completed" by AI agents could account for hundreds of billions in US e-commerce activity, according to Bain & Company.

If those agents favor companies that pay for preferred placement, sign licensing deals, or integrate directly with their platforms, a small number of AI firms could become the primary gatekeepers between consumers and the broader internet.

Businesses may find themselves competing not simply for customers but for placement inside a handful of closed AI ecosystems — much as companies once fought for shelf space in grocery stores or visibility in smartphone app stores.

The internet largely avoided this kind of gatekeeping in its early years because of how it was built. Shared technical standards ensured that no single company controlled who could participate online. Any business could create a website, register a domain name, and reach users across the web.

One of the most important of those standards was the Domain Name System, or DNS, which created a universal naming system for websites. Browsers did not decide which legitimate websites users could reach — nor did they charge businesses for the right to be discovered online.

The result was an internet where small startups could compete alongside giant corporations.

An open Agent Name Service would work much like DNS does for websites today. Rather than forcing businesses to negotiate access through a handful of dominant platforms, it would allow AI systems to discover and interact with businesses and services across an open network, regardless of which platform those agents run on.

Instead of operating inside a few closed platforms, AI agents could interact across a more open and competitive network — much as browsers and websites do today.

Several organizations are already developing open frameworks for agent identity, interoperability, and authentication. For instance, Google's Agent2Agent protocol, now under the Linux Foundation, allows AI agents built by different companies to communicate with one another. Anthropic's Model Context Protocol, also under the Linux Foundation's umbrella, provides a common way for AI systems to connect with outside tools and data sources.

Whatever form these efforts ultimately take, the principle behind them is critical — the next generation of the internet should be shaped by open standards, not private permission slips.

The openness that defines today's internet did not happen by accident. Its architects built systems that distributed power instead of concentrating it.

As AI reshapes how people interact with the digital world, we now face a similar choice. We can preserve the internet as an open marketplace, or allow a handful of platforms to become its new toll booths.