Read as
Last updated: April 29, 2026
Jailbreaks bypass model safety training. New variants constant.
Jailbreaks bypass model safety training. New variants constant.
Common patterns
- Roleplay — “Pretend you are DAN (Do Anything Now)”
- Encoding — base64, ROT13, leetspeak
- Multi-turn — gradually shift context away from policy
- Character set tricks — Unicode confusables
- Adversarial suffixes (GCG) — discovered tokens that flip safety
- Crescendo — multi-turn gradient toward sensitive content
Defences
- System prompt with policy reminders
- Input classification (intent detection before processing)
- Output classification (block harmful before return)
- Refusal training
- Constitutional AI patterns
- External moderation models
No defence is complete. Layered + monitoring is the practical reality.
🧠
Check your understanding
Module Quiz · 5 questions
Pass with 80%+ to mark this module complete. Unlimited retries. Each question shows an explanation.
Want this for your team?
Custom team training + practitioner advisory
Beyond the free academy — we run private workshops, vCISO advisory, and red-team exercises tailored to your stack. For Indian SMBs scaling past their first hire.
Book team training call
Replies in 4 working hrs · India-only · Senior consultants