Every time a new technology appears, the pattern is usually the same.

Businesses adopt it because they see opportunity.

Marketers squeeze every possible use out of it.

Sooner or later, criminals start experimenting with it too.

The internet went through that cycle.

Crypto did as well.

There isn't much reason to think AI will somehow avoid it.

For the last couple of years, most conversations around AI have focused on the upside.

Write code faster.

Answer customers faster.

Research faster.

Produce more with fewer people.

None of that is wrong.

But technologies rarely stay in one lane.

If AI can help someone build software, analyse documents or automate repetitive work, it can also help someone run scams more efficiently or automate parts of an attack that used to require actual manpower.

The phishing email written by AI isn't really the story anymore.

We've been talking about that for years.

The more interesting shift is what happens when attackers stop using AI occasionally and start relying on it the same way companies rely on junior staff.

Because it doesn't care if the task is boring enough to make a human quit after two weeks.

Cybercrime has always had practical limits.

A lot of attacks never happened because putting them together took too much work or required skills that were difficult to find.

Sometimes there simply wasn't enough money at the end of the process to justify the effort.

That matters more than people think.

The internet is full of ideas that are technically possible but economically pointless.

If AI lowers the cost of running an operation, some of those ideas stop being pointless.

Security researchers understand this perfectly well which is why the discussion has started changing.

Cybercrime was always a business before it was anything else.

And businesses have always chased efficiency.

If attacks become cheaper to run, easier to automate and easier to scale, somebody is eventually going to take advantage of that.

Almost every major technology eventually found a role inside cybercrime.

There is very little evidence that AI will somehow be the first exception.