Social media had a 20-year head start. Will AI be different?

By Donya · 11 October 2026

I asked a question today that sent me down a rabbit hole.

When social media first arrived, it was pretty lightly regulated. So how long did it actually take for dedicated safety rules to catch up?

Using Facebook as a starting point, the answer was about twenty years. That immediately made me wonder where AI sits in comparison, and what we could learn from the way we handled social media.

The twenty year gap

Facebook launched in 2004. The UK’s Online Safety Act became law in 2023, and key duties requiring platforms to tackle illegal content became enforceable in March 2025. That is 21 years between Facebook’s launch and that particular regulatory milestone. ^[1][2]

Facebook was not the first social network, and social media was not operating outside the law before 2025. Existing legal requirements and voluntary initiatives already applied. The government’s 2019 Online Harms White Paper described a fragmented approach that had not done enough to protect users. ^[3]

The Online Safety Act introduced a broader framework placing specific responsibilities on platforms to assess and reduce risks, tackle illegal content and protect children, with Ofcom overseeing compliance. ^[2]

What caught my attention was how much time had passed. By the time those duties arrived, social media was already woven into the way we communicated, worked and ran businesses.

Where AI sits in comparison

ChatGPT launched on 30 November 2022. As I write this in October 2026, we are less than four years into this latest wave of widely accessible generative AI. ^[4]

The EU already has dedicated AI legislation. Its AI Act entered into force in August 2024, with requirements introduced in stages. Bans on certain practices began applying in February 2025, followed by rules for general-purpose AI models in August 2025. Further provisions, including transparency requirements, applied in August 2026, while some high-risk requirements have later deadlines. ^[5]

There is an important detail in that comparison: the European Commission first proposed the legislation in April 2021, before ChatGPT launched. AI has a much longer history, and work on its regulation was already underway. ^[6]

Here in the UK, the approach is different. The government continues to rely mainly on existing regulators and laws, depending on how AI is used. These include data protection, equality and product safety rules, alongside requirements in particular industries. ^[7]

So the timeline is interesting, but it is not a like-for-like measure of how quickly governments reacted. We are comparing different technologies, jurisdictions and legal milestones. What it does show is that dedicated AI rules are already arriving during this consumer AI wave.

What we could learn from social media

The question I keep coming back to is how we learn about a technology while it becomes part of everyday life.

It is easy to see what a new tool lets us do. Understanding how it works, where its limits are and what responsibility comes with using it takes more effort. For me, that learning deserves to happen alongside adoption.

Legislation is part of the picture. So is helping people understand the tools they are using. Having a law in place and knowing whether it works well in practice are different things.

I’m excited about AI. I use it, I’m studying it, and I can see how useful it can be. I want more people and businesses to understand those possibilities, and to feel confident asking questions about them.

Looking at social media’s timeline has given me another question to explore. We already have experience of a technology becoming deeply embedded in our lives while the rules around it continue to develop.

What should we learn from social media while we still have the chance to do things differently?