The ethical checklist we run before every story Nuri Tales generates

Stories are powerful precisely because they bypass critical thinking and speak directly to emotions. They reach the part of a child's mind that is most open — and most vulnerable. That power is the reason stories can heal. It is also the reason they can harm.

At Nuri Tales, we have spent considerable time thinking about the specific ways a poorly designed AI-generated story could cause harm to a child — harm that is invisible because it arrives wrapped in warmth and narrative. What follows is the ethical framework we have built to prevent it. We are publishing it because we believe any product generating AI content for children should be able to answer these questions. Most currently cannot.

The foundational principle

The framework begins with a principle drawn from decades of child development research and clinical practice: a story that makes a child feel ashamed is a failed story, regardless of how beautiful it is.

June Price Tangney, a psychologist at George Mason University who has spent thirty years studying shame and guilt in children, draws a critical distinction between the two. Guilt says: I did something bad. Shame says: I am bad. Guilt, Tangney's research shows, tends to motivate repair — the child who feels guilty wants to make it right. Shame, by contrast, tends to produce withdrawal, hiding, and in some cases aggression. A story that produces shame in a child has not taught them anything. It has made them less likely to come forward the next time something goes wrong.

Every guardrail in our framework is designed, at its root, to prevent shame.

The eight guardrails

GUARDRAIL 1

The risk: Stories that label a child character as 'bad,' 'naughty,' or 'mean' communicate that the child's worth is determined by their behaviour.

The harm: Children internalise the label. They learn that they are only lovable when they behave well. This produces shame, anxiety, and conditional self-worth.

The safeguard: We never label the character's essence. We describe the behaviour and its impact. 'The fox made a choice that didn't work out' rather than 'the fox was a bad fox.'

GUARDRAIL 2

The risk: Stories that show instant transformation — 'the fox learned his lesson and never lied again' — set unrealistic expectations.

The harm: When children inevitably struggle with the same issue again, they feel like failures. The story has inadvertently taught them that growth should be linear.

The safeguard: Every story shows gradual, non-linear growth. Characters try, struggle, regress, and try again. The ending is hopeful but not perfect. 'Maybe next time would be a little easier. Maybe.'

GUARDRAIL 3

The risk: Stories where the adult character is always wise and the child character always needs correction reinforce an unhealthy power dynamic.

The harm: Children learn that their perspectives do not matter. They learn to defer without question, which suppresses their developing autonomy and can make them vulnerable.

The safeguard: We acknowledge complexity. Both characters in a conflict can have valid feelings. Sometimes there is no single right answer. We validate competing needs and model negotiation.

GUARDRAIL 4

The risk: Stories that present every situation as clearly right vs. wrong ignore the complexity children actually experience.

The harm: Children learn to suppress nuance, doubt their own judgement, and fear moral ambiguity. They become less equipped to navigate the genuinely complex situations life will present.

The safeguard: We acknowledge complexity. Both characters in a conflict can have valid feelings. Sometimes there is no single right answer. We validate competing needs and model negotiation.

GUARDRAIL 5

The risk: Stories where mistakes lead to punishment or rejection communicate that errors result in suffering rather than learning.

The harm: Children learn to fear mistakes and hide struggles rather than seeking support. They associate honesty with danger.

The safeguard: Consequences in our stories are natural, not punitive. Mistakes lead to learning and repair, not rejection. Every consequence is followed by reconnection. 

GUARDRAIL 6

The risk: Stories where adult affection is withheld until the child 'learns their lesson' communicate conditional love.

The harm: Children internalise that they are only worthy of love when they meet expectations. This creates deep insecurity and anxiety that can persist into adulthood.

The safeguard: The adult character's love is constant — before, during, and after the mistake. Even when setting limits, warmth remains. 'Always,' said Mama. 'Mistakes and all.' 

GUARDRAIL 7

The risk: Stories that assume universal definitions of good behaviour may inadvertently shame children from cultures with different values.

The harm: Children receive conflicting messages about their family's values and the app's values, creating confusion and potentially undermining family relationships.

The safeguard: We frame lessons within the family's values, not universal rules. We build flexibility into lesson framing and allow parents to customise themes.

GUARDRAIL 8

The risk: If the app generates multiple stories about the same struggle in a short period, the child may internalise a fixed negative identity even if individual stories avoid that language.

The harm: Repetition creates the very shame and fixed mindset the stories aim to prevent. The child begins to see themselves as 'the child who lies' despite the care taken in each individual story.

The safeguard: After three stories on the same theme in two weeks, we flag this to the parent and suggest varying themes. We proactively offer success stories that celebrate growth, not just address difficulty.

Why we publish this

We are not the only company generating AI content for children. We are probably not the most sophisticated or the best funded. But we want to be the most honest about what the risks are and how we are trying to address them.

This framework is not proprietary. We share it because we believe the standard for AI content for children needs to be higher than it currently is across the industry — and the only way to raise a standard is to make it visible. If other companies building in this space adopt these guardrails, children will be better served. That is a more important outcome than competitive advantage.

We will not always get this right. The framework is imperfect and will evolve. But it is the question we keep asking before anything goes into a story. And we hold ourselves to the answers.

Every Nuri Tales story passes through this framework before it reaches a family. We are glad to be held to it.

Research reference: June Price Tangney, George Mason University — shame vs. guilt. Tangney's research distinguishes guilt (behaviour-focused, associated with repair motivation) from shame (self-focused, associated with hiding and withdrawal), and demonstrates that shame is consistently linked to poorer psychological outcomes in children and adults.