Right now, the tools are still generous. Free tiers are still usable. Experiments are still cheap enough that y'all can learn by doing, instead of waiting for permission, a budget, or a procurement department to sign off. That window is not going to stay open forever.

Governments are moving faster than you think

On 13 December 2024, the UK government introduced the Digital Markets, Competition and Consumers Act. It gave regulators new powers to fine tech companies up to 10% of global revenue for anti-competitive behavior and to block mergers outright if they threaten fair competition.

A week later, on 21 December 2024, the EU passed the AI Act. It went into effect on 1 February 2025. The law imposes fines of up to €35 million or 7% of global revenue for violations. That is not a warning shot. That is enforcement with teeth.

Italy is another good example. At the end of December 2025, the Italian competition authority ordered Meta to suspend WhatsApp conditions that would have blocked rival chatbots like ChatGPT, Claude, and Perplexity on the platform. Brazil's regulator did the same in January 2026. The European Commission opened its own parallel investigation. This is not a niche legal fight. It is a clear signal that governments are willing to step in when they think access, competition, or control is being squeezed too hard by a single platform.

We have seen this story before with the internet itself. The early days felt open, messy, and under-regulated. Then came privacy laws, content moderation, platform gatekeeping, and a long list of standards that made the internet safer in some ways and harder to navigate in others.

AI is following the same path. Faster.

That does not mean regulation is bad. It means open, frictionless access usually does not stay open and frictionless forever. The people who benefit most in those shifts are the ones who learned the system early, before the rules got heavier and the costs got higher.

Company limits are tightening, too

The corporate side is moving in the same direction. OpenAI announced ads inside ChatGPT on 16 January 2026 and started serving them on 9 February 2026 to free and Go-tier users in the US. By late March, the company was already expanding the pilot into Canada, Australia, and New Zealand. Plus, Pro, Business, Enterprise, and Education plans stay ad-free for now. The trajectory is the obvious one. Free tiers shrink. Ads creep in. Paywalls grow. The most useful features quietly move behind a subscription.

This is not unique to OpenAI. Free tiers across the industry are getting tighter. Rate limits are getting stricter. Features that felt like default a year ago are turning into upgrades. What used to feel like a generous playground is becoming a metered service. More quotas. More message caps. More model-specific restrictions. More "you can try this, but only so much." The everyday user is going to end up with a more limited, more segmented, and more expensive experience than they have today.

What to do now

The advantage does not belong to the people who wait. It belongs to the people who start now. Not just learning a tool, but learning how to use it without letting it replace the hard, messy, irreplaceable work of showing up in real life. If the next few years bring more polished companions, more convincing voices, more perfect validation, then the best time to get good at real connection was yesterday.

The second-best time is right now.

— Eduardo Cestaro