The manual
The full framework in practical form — guardrail prompts, decision tables, boundary templates and a chapter on children and family safety.
Home › SAFE Framework
Why a framework at all
Rules age badly. Judgment travels.
A list of things not to do goes out of date the moment the tools change — and they change monthly. What does not go out of date is a way of thinking you can apply to a tool that did not exist when you learned it.
SAFE is deliberately not a set of prohibitions. It is four questions you learn to ask, in order, until asking them becomes automatic. A parent, a nurse, a student and a school district all run the same four questions — only the stakes and the context change.
The four questions underneath the four pillars
- Structure — What am I actually trying to do here, and what are my limits?
- Accelerate — Is this a task where AI genuinely extends what I can do?
- Filter — Is this output any good, and can I tell?
- Empower — Am I still the one deciding, and am I accountable for it?
Putting it to work
Where the framework shows up
SAFE is not a page you read once. It runs underneath everything we make.
Your assessment plan
The assessment ends by mapping your results back to these four pillars, so the plan you get is built on the same method.
Training and curriculum
Institutional programs teach the same four pillars, so a whole staff or student body ends up with one shared language for responsible use.
About this framework. The SAFE AI Use Framework™ is the original work of the AI
User Safety Initiative. Where our materials reference
international AI literacy, ethics or risk-management guidance, those are sources we have drawn on —
not endorsements of our programs, and not accreditation or certification by any of those bodies.
Four questions. Every time. Until it's automatic.
That is the whole method. The manual shows you what it looks like in the situations you actually run into.