World Wide Web Day 2025: The Fight to Fix the Internet

On 6 August 1991, Sir Tim Berners-Lee published the first website from his office at CERN, launching what would become the World Wide Web.

The British computer scientist’s creation was built on principles of openness and decentralisation: a system where information could be freely shared.

Deliberately choosing not to patent his invention, believing the Web should belong to everyone, his original vision emphasised user empowerment: individuals would control their own data whilst accessing a vast network of interconnected information.

Thirty-four years later, that vision looks different.

A handful of technology giants now dominate the internet, harvesting personal data on an unprecedented scale.

Users routinely surrender intimate details of their lives – browsing habits, location data, social connections, purchasing behaviour and more – often without understanding the full implications.

How big tech changed the web’s original vision

The transformation from Berners-Lee’s decentralised Web to today's platform-dominated internet didn’t happen overnight.

The shift accelerated in the 2000s as companies like Google and Facebook built business models around data collection and targeted advertising.

What began as ‘free’ services funded by advertising evolved into sophisticated surveillance systems, tracking behaviour patterns, social connections and even offline activities through location data and purchase tracking.

The result is an internet where users have become the product rather than the customer.

Personal data is extracted, processed and monetised by companies whose algorithms increasingly determine what information people see and which choices they’re offered.

Why AI raises the stakes for data control

AI has fundamentally changed how personal data can be used.

Modern AI systems can synthesise vast amounts of information to reveal intimate details about individuals’ thoughts, preferences, and behaviours that were never explicitly shared.

These systems can predict actions, generate personalised content and make decisions based on patterns in data that humans might never notice.

The scale of data processing involved dwarfs anything previously possible, turning every digital interaction into potential training material for AI models.

“We’re at a critical point where, if we don’t watch it, our data becomes consumed by LLMs,” said John Bruce, who co-founded Inrupt with Berners-Lee, at London Tech Week 2025.

“As a consequence, we'll become their servants. They’ll make decisions for us too easily.”

Alternative models for data ownership

Several initiatives are exploring ways to restore user control over personal data.

These range from technical solutions like decentralised identity systems to regulatory approaches such as the EU’s Digital Services Act.

One approach involves ‘data pods’: digital spaces where individuals store and control their own information, granting specific permissions for access whilst maintaining ultimate ownership.

This could allow people to benefit from AI services without surrendering permanent control over their personal information.

Financial services companies are investigating customer data sovereignty models, while media organisations are considering how to maintain user trust whilst delivering personalised services. Government interest is also growing across Europe, the US and Asia.

However, changing entrenched systems proves difficult.

The current model generates enormous profits for tech companies and has become deeply embedded in digital infrastructure. Building alternatives requires both technical innovation and shifts in user behaviour.

The challenge of timing

The window for implementing alternative approaches may be narrowing as current systems become more entrenched. Network effects and switching costs make it increasingly difficult for users to move away from established platforms.

This challenge is compounded by the rapid pace of technological change. When Berners-Lee first wrote about building an AI platform six or seven years ago, John says they expected a 20-year development timeline. “Here we are, seven years later, and we built the AI platform because that’s what the technology can do these days.”

The speed of development means there is little time for course correction. Yet the greatest challenge may not be technological, but one of perception.

The web has become so synonymous with a handful of dominant services that its original, more radical potential is easily forgotten.

It’s a challenge in understanding that Bruce believes is the first hurdle to clear.

“People sort of get it and sort of kind of don’t, but give yourself the opportunity, give yourself the chance to get it,” he urges.

“Go on the Web, as invented by Tim, and experience it a bit more.”