Beschrijving
Ever wonder why the AI tools at work feel so locked down — features switched off, access you can't get, a chatbot that won't touch half your data? This is why. You've been warned about what you type into AI; the bigger risk is what AI reads — from an email, a web page, a document — without you ever seeing it. In under nine minutes, Simon Allardice unpacks prompt injection (the threat OWASP ranks number one), how it differs from jailbreaking, why it can't simply be patched, and what you can actually do to limit your exposure.
Want to go deeper on responsible AI and security? Check out Udacity's Responsible AI Nanodegree: https://www.udacity.com/course/responsible-ai--nd700
#promptinjection #jailbreak #responsibleai #artificialintelligence #cybersecurity #techexplained