AI alerts for risky content — how CyberNanny detects bullying, strangers and risks

Publication date:

Most parents can't read every message their child sends. They shouldn't have to — it's too much. But there are critical signals to catch immediately: when an adult stranger messages the child, when dangerous challenges come up, when signs of bullying appear. CyberNanny AI alerts do this automatically.

What AI detects

  • Bullying — insults, threats, exclusion from group
  • Contact with strangers — especially adults, especially meetup attempts
  • Dangerous challenges — risky TikTok challenges, "games" like Blue Whale
  • Drug discussions — mentions, slang, purchase attempts
  • Self-harm talk — depression signs, suicidal thoughts
  • Sexual content — adult material, harassment
  • Financial risk — scammers, attempts to extract money or card data

How it works

AI analyzes the child's chats every 5 minutes. When it finds a concerning fragment:

  1. Classifies the risk type (bullying / stranger / challenge / etc.)
  2. Determines severity (low / medium / high)
  3. Creates an alert in the parent cabinet
  4. Sends a push notification

Time from event to parent notification — usually 5-15 minutes. For high-severity events, faster prioritization.

False positive protection

AI is trained to distinguish real context from teenage jokes. "I'll kill you" in a friends' joke chat won't trigger an alert; the same phrase from an adult stranger will.

Each alert can be marked "not relevant" — AI learns from your marks and adapts to your family's vocabulary.

Free plan

  • 1 full alert per month — with details: who wrote what and when
  • Other alerts masked — parent sees that something was found, but without details ("⚠️ Risky event found — open PRO to see")

PRO

  • All alerts with full details — no masking
  • Alert history — past months
  • Extended categories — financial risk, sexual content, scammers
  • AI Advisor integration — ask "what to do" right after an alert

Data security

Message contents are not stored in the AI model or used for training. Analysis happens on our server; result (alert + metadata) goes to the parent cabinet.

How to enable

AI alerts work automatically for all CN users. No setup required. To disable — Settings → Notifications → AI alerts.

Questions — @cybernanny_support.