25/08/2026
I’ve just finished a small internal review of how I actually use AI in NetSwifter work.
Not because NetSwifter is becoming an “AI company”. It isn’t.
AI is simply one of the tools I use for things like website analysis, Search Console data, SEO, coding assistance, content development and quality checks.
The new EU AI Act transparency rules gave me a good reason to look at the process properly and ask a few practical questions:
* Do AI-assisted articles need to be labelled?
* What about realistic AI-generated images?
* How should client data be handled?
* Does AI-assisted code create new obligations?
* Where should human review remain non-negotiable?
The conclusion was quite simple:
**Human-led, AI-assisted.**
**Minimum necessary data.**
**Never fabricate evidence.**
I’ve written up what I learned, including the changes I made to the way NetSwifter handles content, images and client work:
https://netswifter.com/ai-generated-content-eu-ai-act/
I also published a short **NetSwifter AI Policy** explaining our approach:
https://netswifter.com/privacy-policy
The interesting part for me wasn’t finding reasons to use less AI.
It was defining where AI genuinely improves the work, where it needs tighter controls, and where professional judgement still has to lead.