AI Leakage Alerts
Find out when your AI tools change the rules — from us, not from a breach
AI vendors quietly change what they train on, how long they keep your data, and who can see it. Most small businesses never notice. We watch the tools you actually use and send you a plain-English heads-up with one thing that matters: what to do about it.
Get the alerts — free
Plain-English. Only when something actually changes. One click to leave, any time. We use double opt-in, so you'll get a quick confirmation email to click — then you're on.
Trouble with the form above? Subscribe here instead →
No spam · Unsubscribe in one click · Independent (no vendor pays us)
What you'll actually get4 things
Default changes that bite
When a tool flips to training on your data by default, or moves the opt-out — the quiet changes nobody emails you about.
Incidents that matter
Real, sourced breaches and vulnerabilities in tools small businesses use — not generic "AI is scary" noise.
The action, not just the news
Every alert ends with the specific step: the setting to change, or the tier to move to. Decisions, not headlines.
Good news too
When a tool changes in the safer direction, we'll tell you. We're not here to scare you into anything.
See one before you subscribeA real recent alert
AI Leakage AlertsSample
🔴GitHub Copilot
Now trains on your code by default on Free/Pro · Our rating: Free/Pro 4 of 5 · Business/Enterprise 2 of 5 (unaffected)
What changedAs of April 2026, GitHub Copilot's Free, Pro and Pro+ plans default to allowing your code to be used to improve the product. It used to be opt-in. It's now opt-out — on unless you turn it off.
Who it affectsAnyone on Free, Pro or Pro+. Business and Enterprise plans are not affected — they exclude training by contract.
What to doIf you write any proprietary or client code: turn it off in your GitHub Copilot settings, or move to a Business seat (which excludes training by contract). Until then, keep confidential code out of Free/Pro Copilot.
Why we're sureGitHub's own Copilot privacy documentation, change effective April 2026.
That's one item. A typical issue has 2–5, ordered by how urgently they matter.
The honest small printFAQ
How often will you email me?
Monthly to start, with the occasional same-day heads-up if something genuinely urgent breaks. We'd rather skip a month than pad your inbox.
Is it really free? What's the catch?
Free, no catch. Later we'll offer paid tools for businesses that want more depth — but the alerts stay free. We're independent: no vendor pays us for a good rating or a mention.
What do you do with my email?
Send you the alerts. That's it. Double opt-in, one-click unsubscribe, never sold or shared. We're an AI
data-responsibility brand — running a clean list is the whole point.
How we work →Will this be useful if I'm not technical?
That's exactly who it's for. Plain English, no jargon, written for the owner-operator who has no IT department — not for security professionals.
AI Leakage Alerts is independent research for small businesses — we are not a vendor and nobody pays for a rating. Alerts are general guidance, not legal advice; obligations under the NZ Privacy Act 2020 or the Australian Privacy Act may need professional advice. You can unsubscribe at any time with one click.