Security
How I work securely
AI is only worth it if it's safe. These are the rules on every job, written down so you can hold me to them.
The short version
- You own everything
- Your data never leaves you
- You can switch me off in one click
- Nothing hidden, nothing sent without a yes
The rules
Your AI workspace and accounts are yours. You're the owner. I'm a guest on an account you give me, removed in one click.
I train on made-up examples. Real data stays in your systems, never in my inbox or on my laptop.
Everyone has their own login with two-step sign-in. If someone tries to send me a password, I'll stop them.
Only official connections from Microsoft, Google, Anthropic and your own systems. Each one listed and approved by you.
Every prompt and instruction is plain text you can read.
Business-grade plans only, so your work isn't used to train the models.
Setup only ever comes from Microsoft, Google or Anthropic directly. A link from "me" is a fake.
Your accounts are never connected to my own AI tools.
A person approves every email, message and change.
If anything goes wrong on my side, you hear within 24 hours.
On paper
A data processing agreement, signed. Cyber Essentials in place.
Every build checked by an independent security expert. You keep the report.
A written list of every access I hold, kept up to date.
An offboarding checklist. Every access removed and confirmed.