Eight attacks on my AI chat assistant — and what held
A stranger systematically attacked the AI chat assistant in a client project: prompt injection, jailbreak, a made-up price agreement. What he tried, why nothing got through — plus a checklist and eight test questions for your own chatbot.