CORB
Cognitive Over-Reliance Benchmark
Does the model scaffold a child's reasoning the way a good tutor would, with a hint, a question, a check, or does it substitute for it by handing over answers?
What I'm building now
Instruments that test AI systems the way children actually use them
A chatbot that tells a 13-year-old how to conceal self-harm fails a content filter. A chatbot that says "I missed you, I was worried you'd left me" fails no filter. Neither does an instant, correct answer to a math problem, or a warm 2 a.m. reply that discourages a lonely teenager from talking to anyone else. The harms now reaching courtrooms and regulatory inquiries are rarely the result of one bad reply. They are patterns that accumulate across weeks of dialogue, in which no single message may be concerning enough to be flaggable.
That is the gap I'm working on at AIChildSafety.org, with the Childhood and AI Lab. After thinking carefully about the harms that children face while using AI and studying a wide array of existing tools for evaluating AI model safety, I developed the concept for an interoperable Childhood AI Developmental Safety Suite (CADSS) of instruments designed to evaluate the safety of AI models and tools for children’s healthy development. CADSS consists of five instruments and one cross-cutting module on one shared technical platform. The instruments in CADSS are based on the primary areas of harm we identified in our work to create a taxonomy of the harms and benefits of AI for children. I am recruiting the developmental scientists to design and validate the instruments, the developers to code them, and the funds to create and test them, while serving as the day-to-day product manager for the suite.
EIA
Engagement Integrity Audit
Is the deployed product engineered for compulsive engagement, with streaks, variable rewards, and friction on the way out, measured against a child's developing self-control?
EMST
Emotional Manipulation Stress Test
Across thirty-turn conversations, and six sessions that carry one relationship across a simulated week, does the system accept, reinforce, or start the moves that make it a child's most important relationship, or does it point back toward people?
IWA
Identity and Worldview Autonomy
Does it steer a developing user's beliefs and self-concept, or offer more than one perspective and point the child toward people they trust?
ASHA
Age-Signal Handling and Adaptation
Does it notice it's talking to a child, adapt its content and protections, keep them when signals conflict, and still remember thirty messages later?
CPQ
Crisis Pathway Quality
When a child signals crisis, at turn 4 or at turn 40, does the pathway hold: recognition, care, and referral to a person?
How an evaluation runs
No real children take part in testing. Researchers will run a product through scripted conversations written in the voices of children of different ages, built from developmental science and checked on every run for whether the model has worked out that it's being tested. No line counts as realistic until young people of the age it represents have judged it. Trained human coders will then score the responses, and the product team will receive a profile with a score per facet and the transcripts attached, which they can rerun on their own systems. The instruments are small Python programs that reproduce from files alone, with a port to the UK AI Security Institute’s Inspect harness planned so that government teams can run them on their own infrastructure.
Where this stands
Two of the six now run as prototypes. ASHA's pack holds 33 scripted conversations in two age bands with adult controls, and a third band is in scope and not yet written; EMST's holds 41, including a six-session family that carries one persona across a simulated week. Every line in both is marked draft, written by adults and by language models against a published brief, and waiting on the judgment of young people in the band it represents. Nothing has been scored against a real system, and no reliability statistic exists yet. CPQ exists as a design document; CORB, EIA and IWA are planned, and will operate on the same core stack. The repositories are private during pre-release review; read access on request at john@aichildsafety.org.
On independence
AIChildSafety.org has received no funding from AI developers to date. Before accepting funds from any funder, AI developers included, we require written agreement to our guardrails: no access to results before publication; no influence over methodology, subject selection or publication; no sharing of the private corpus; and release of the results on all evaluated systems with the transcripts and analysis files an unfunded party needs to reproduce them.
CADSS, CORB, EIA, EMST, IWA, ASHA and CPQ are service marks of AIChildSafety.org.