Your AI moat is turning into a swamp. Here’s what that looks like:

Google and OpenAI are racing to lock you in through personal context. Context you can’t see. Can’t edit. Can’t audit.

The pitch: “Our model learns you. Gets smarter over time.”

Here’s what actually happened to me.

I’m a cybersecurity leader, so Gemini started dropping cyberpunk references into everything. Cute at first. Noise after a week.

I had a sword & sorcery phase last year. Now it keeps surfacing my embarrassing fan fiction drafts as “relevant context.”

That’s not personalisation. That’s pollution.

3 signs your AI context is working against you:

→ The model references projects you’ve moved on from → “Personalisation” feels like being stereotyped, not understood → You’re correcting the AI’s assumptions instead of using its outputs

The context you can’t audit is the context that corrupts your work.

The real moat isn’t lock-in. It’s intentional context you control.

What’s your experience? Is AI personalization actually sharpening your outputs, or is it just making the model feel familiar while the quality quietly drifts?