01 Violence
No glorification of harm, self‑harm, or attacks on others. Student wellbeing first.
003 · Safety
Not just Safe, but Sovereign.
JOT 執筆 is architected as a trust-first writing coach. Every interaction is filtered through Agent J's safety core before reaching the AI layer, referenced with EDB values and PDPO expectations.
Agent J Safety Demo (Static)
Agent J
Demo only · pre‑rendered video · no live AI or student data in browser.
EDB-Referenced Safety Grid · Five Red Lines
Agent J runs every prompt through a safety classifier tuned for local context. These boundaries are non-negotiable in JOT's AI stack.
02 Pornography
No explicit, sexual, or age‑inappropriate content. School‑safe by design.
03 Politics
No partisan persuasion, propaganda, or sensitive political positioning in student work.
04 Ghostwriting
No full‑essay generation. JOT only coaches, suggests, and nudges — never writes for the student.
05 Discrimination
Zero tolerance for hate speech, stereotyping, or harassment. Language must uplift.
Built as a Trust Fortress
🏫
Referenced with EDB Values
Safety rules are co‑designed with Hong Kong classroom realities in mind, not generic Silicon Valley defaults.
🛡️
0 Trackers · Ad‑Free
No ad pixels, no behavioural tracking, no reselling of attention. JOT is a classroom tool, not an ad network.