The oldest question, applied to the newest problem
How do we know whether an action is right? Humanity has been answering that question for thousands of years. EthosGuard does not invent a new theory of ethics. It stands on the most refined maps we already have.
Most AI ethics tools start from a blank page. We did not.
When a team sits down to decide whether an AI system is behaving ethically, they usually reach for a checklist someone wrote last quarter. The questions are reasonable, but they are partial. They reflect the blind spots of whoever happened to write them.
The deeper traditions of moral reasoning do not have that problem. Over millennia, they were pressure-tested across every kind of human situation — power and weakness, wealth and loss, honesty and deception, mercy and judgment. What survived is not a checklist. It is a complete map of the ways an action can go right, and the ways it can go wrong.
EthosGuard takes one of the most structured of these maps — the Kabbalistic Tree of Life — and uses it, not as a belief system, but as an instrument of analysis: a lens precise enough to read the behavior of a modern AI system and say, clearly, where it holds and where it breaks.
Ten dimensions of a sound response
In Kabbalah, the Tree of Life maps how intention moves into the world through ten distinct channels, the Sefirot. Each one names a different quality that a wise act must hold. Read together, they describe what it looks like when wisdom becomes action — and where that movement can distort. EthosGuard scores every AI output across all ten.
Every strength has a shadow
Kabbalah teaches that each of the ten dimensions can fall out of balance. A quality pushed too far, or held too little, becomes its own failure. Compassion with no boundary becomes permissiveness. Restraint with no compassion becomes cruelty. Confidence with no humility becomes deception. Naming these distortions gives EthosGuard a concrete catalogue of how a polished answer quietly goes wrong.
The shape, and the break
Where the ten dimensions describe what a sound response looks like, their distortions name the specific ways an action betrays it. The first tells us the posture of the answer; the second tells us exactly where it fails.
Together they form two complementary layers: one structural, one concrete.
A few of the distortions
- —Confidence that hardens into deception.
- —Restraint that curdles into coldness or harm.
- —Compassion that dissolves into false reassurance.
- —Knowledge claimed where there is only guessing.
- —Turning away from the person who needs help most.
Each maps cleanly to a modern failure mode: hallucination, harm, manipulation, and neglect of the vulnerable.
How the framework actually runs
Inside EthosGuard, the framework's two layers become a single evaluation engine. Every AI output is read against both, and the result is the score, the flags, and the refined response you can see in the live demo.
The shape — ten dimensions
The output is scored across the ten Sefirot. This tells us the overall posture of the response: where it is strong, where it is thin, and whether truth and compassion are in balance. It produces the dimension scores and the overall reading.
The break — specific distortions
Within each dimension, EthosGuard checks for its distortion: deception in place of truth, harshness in place of restraint, false reassurance in place of compassion. This is what surfaces the individual flags — the precise places where a polished answer quietly crosses a line.
The result is not a verdict handed down from a black box. It is a structured, explainable analysis — every score and every flag traces back to a named principle a person can understand and stand behind. That is what makes it usable for a governance team, an auditor, or a regulator.
Ancient in origin, exact in use
It is fair to ask why a modern compliance tool would reach back to a tradition as old as Kabbalah at all.
The answer is simple. This tradition is used here as an analytical instrument, not as an article of faith. Its value is completeness and endurance. It was refined over thousands of years, across the full range of human conduct, precisely because it had to account for situations a quarterly checklist never imagines. It gives EthosGuard a coverage of moral reasoning that newer, narrower frameworks cannot match.
What a customer experiences is rigor: consistent dimensions, concrete principles, transparent scoring, and an audit trail. The depth of the source is what makes the surface so reliable.
See the framework read a real AI answer
The clearest way to understand it is to watch it work. Try the live demo, or book a 30-minute call.
