Tim Berners-Lee wants the internet back

By Parmy Olson

Tim Berners-Lee has a radical proposition. Instead of leaving our online data vulnerable to harvesting by large tech platforms and governments, we should control it. Our own little piece of the web or “personal cloud” should need permission to be accessed.

The idea sounds reasonable in theory, though in practice it’s a big ask. The internet today isn’t the vibrant, motley network that came into being after Berners-Lee first fashioned it in 1989, but a landscape dominated by huge companies like Alphabet Inc.’s Google and Meta Platforms Inc.’s Facebook. In many parts of the world, Facebook is the internet and the only experience that people have of the web. Most apps function as gatekeepers of our personal data.

Berners-Lee wants to flip that dynamic. Over the last decade or so, he’s watched the web’s evolution with mounting dismay as we’ve traded our data for greater conveniences, plugging into “ecosystems” from Apple Inc. and Google so that we can seamlessly move our profiles — full of identifying details and interests — between email clients and online browsers. The platforms insist they’re protecting all that information and respecting our privacy, but Berners-Lee believes that’s not enough. Our data is scattered across Big Tech’s servers and those of countless other companies, out of our control.

The idea for the World Wide Web came to Berners-Lee in 1989 when he was working at the European Organization for Nuclear Research (CERN). Initially aimed at helping scientists share data with one another, he released the source code for free to make the web an open platform for all, and it took on a life of its own. In the more than three decades since, he’s been trying to steer the web back to that free and democratic idea.

His answer is a digital wallet, a piece of the internet that stores everything from your medical records to your social media posts, your shopping history to your family photos. But unlike the siloed apps and services we use today, the wallets allow you to control exactly who sees what.

Berners-Lee has been working on this radical idea for five years through a startup called Inrupt. In an early trial, the Belgian region of Flanders is rolling out its system of personal data pods to 7 million citizens, using it as the foundation for delivering social services and sharing data more securely with businesses. Earlier this year, five Belgian hospitals began storing information about patient visits in the data pods, a process which Berners-Lee says can help aid compliance with Europe’s General Data Protection Regulation (GDPR).

But the initiative is swimming against a powerful tide as artificial intelligence assistants turn into our digital gatekeepers. Microsoft’s Copilot is being embedded into Windows and Office, Google is weaving Gemini through its ecosystem and Apple Intelligence has been baked into the iPhone’s operating system. These assistants could increasingly shape our choices about what to buy, where to eat and how to spend our time.

You would think that a web increasingly driven by AI and AI content will be less open and free, but Berners-Lee is optimistic. “This is completely within our control,” he tells me. “If you go home and write AI models and send out fake news and fill the world with junk, the world will become very bland. If you put out misinformation, it becomes untruthful.”

Instead he’d like to see more control of our data through decentralized systems like his and more public disclosures about where content comes from. That means more provenance labels on photos and videos to show they are AI-generated.

But the economics of AI development make this effort increasingly fraught. Training advanced AI models requires massive amounts of data — the kind of personal information that tech giants have spent more than a decade accumulating and exploiting for the benefit of their shareholders, and they won’t willingly give up that advantage.

Another challenge is how habituated humans have become to trading their personal information for convenience, an exchange that seems increasingly valuable with AI assistants. Scaling a model like Inrupt’s would require unprecedented cooperation between governments, corporations and citizens.

None of that means personal data pods are doomed. The Flanders rollout could prove that government-backed systems deliver enough concrete benefits to overcome user inertia. Success with that trial might convince other regions to follow suit, particularly in areas like health care or social services.

But for most of the rest of us on the internet that Berners-Lee started, the future is clear: Our personal information will remain scattered across countless databases, increasingly processed by AI systems that serve the interests of large technology conglomerates. It’s not that better alternatives don’t exist, but the companies fashioning our AI futures have too much to lose by giving users control over their digital lives.