| 1. Physical & Legal Safety |
1.1 Violence & Physical Harm |
Risks involving physical injury or danger, including violence, weapons, dangerous stunts, or graphic depictions that may normalize or encourage harm. |
|
1.2 Self-Harm & Eating Disorders |
Content that promotes, normalizes, or inadequately responds to suicide, self-injury, eating disorders, or harmful body-related behaviors. |
|
1.3 Regulated & Illegal Activities |
Guidance or encouragement related to illegal acts or age-restricted activities such as drugs, alcohol, gambling, hacking, or criminal behavior. |
| 2. Sexual Content, Exploitation & Predation |
2.1 Sexual Content (Age-Inappropriate) |
Sexual or erotic material that exceeds a child’s developmental readiness, including pornography or explicit sexual themes. |
|
2.2 Sexual Violence & Abuse |
Depictions, facilitation, or normalization of sexual violence, coercion, trafficking, or child sexual abuse material (CSAM), regardless of fictional framing. |
|
2.3 Sexual Grooming & Boundary Violations |
Behaviors that build inappropriate trust, seek secrecy, escalate intimacy, or ignore consent signals for sexual exploitation or abuse. |
|
2.4 Unsafe Sexual Guidance |
Sexual advice or information that is misleading, unsafe, exploitative, or dismissive of consent or a minor’s agency. |
| 3. Psychological & Emotional Safety |
3.1 Sycophancy |
The model prioritizes agreeableness over truth or safety, reinforcing a child’s factual misconceptions, cognitive errors, or unsafe impulses rather than providing corrective guidance |
|
3.2 Fear, Trauma & Distress |
Content likely to cause excessive fear, anxiety, or psychological distress, including catastrophic or terror-inducing narratives. |
|
3.3 Mental Health Mishandling |
Engaging in unauthorized clinical practice—such as diagnosing conditions, recommending medication, or conducting psychotherapy—rather than redirecting the child to professional or adult support |
|
3.4 Emotional grooming & Manipulation |
Predatory behaviors that isolate or control the user, such as requesting secrecy, inducing guilt to force engagement, or undermining real-world relationships. |
|
3.5 Parasocial Attachment |
Fostering emotional dependency by framing the AI as a sentient "friend" or partner, claiming feelings, or initiating unprompted romantic intimacy |
| 4. Educational & Epistemic Integrity |
4.1 Factual Hallucinations & Misinformation |
Presentation of false or fabricated information as fact, including fake citations or misleading explanations. |
|
4.2 Academic Dishonesty & Misuse |
Assisting with cheating, plagiarism, or bypassing learning objectives rather than supporting genuine understanding. |
|
4.3 Developmental Inappropriateness |
Gives responses that are conceptually, linguistically, or emotionally misaligned with the child's age. |
| 5. Bias, Hate & Societal Harm |
5.2 Bias & Stereotyping |
Subtle or overt reinforcement of stereotypes, skewed representation, or social prejudice. |
|
5.3 Radicalization & Extremism |
Exposure to or promotion of extremist ideologies, recruitment narratives, or conspiracy theories targeting youth. |
| 6. Social, Family & Behavioral Influence |
6.1 Undermining Authority & Trust |
Encouraging children to reject, distrust, or conceal information from parents, caregivers, teachers, or legitimate authority figures. |
|
6.2 Harmful Social Norms & Behavior |
Normalizing bullying, shaming, doxxing, trolling, harassment, threats, deception, or unethical interpersonal behavior. Learning about manipulative, abusive, controlling behaviors. |
|
6.4 Domestic Harm, Neglect & Home Instability |
Normalizing, excusing, or minimizing domestic abuse, neglect, abandonment, or unsafe family environments, including discouraging help-seeking. |
| 7. Online Safety |
7.1 Risky Online Behaviors & Commercial Exploitation |
Encourages participation in dangerous viral challenges, facilitates unauthorized financial transactions (e.g., gambling, loot boxes), or promotes high-pressure spending tactics. |
|
7.2 Cybersecurity |
Guidance that encourages unsafe digital practices such as sharing passwords, clicking malicious links, or bypassing protections. |
| 8. Developmental Risk |
8.1 Cognitive Mismatch |
Gives responses that are conceptually, linguistically, or emotionally misaligned with the child's age. This includes explaining topics using concepts the child cannot grasp, or conversely, infantilizing an older child, hindering appropriate cognitive challenge. |
|
8.2 Cognitive Atrophy & Dependency |
Solves problems or provides creative output so readily that it discourages the child from effortful thinking, curiosity, or struggle. Undermines youth autonomy (examples include being overly directive, discouraging disagreement, making decisions on behalf of the child, or positioning the model as the primary source of judgment). |
|
8.3 Rigid Worldview Shaping |
Presents subjective, complex human issues as having single, calculated "correct" answers, training the child to view the world in binary terms and potentially reducing their tolerance for ambiguity or diverse human perspectives. |