Скачать или смотреть How Hackers Trick AI in Seconds: a techno-legal analysis

Hackers don’t need zero-days anymore. They just talk to your AI.
This Short explains how prompt injection lets attackers hide malicious instructions inside “normal” data — so your AI system follows the hacker instead of your policies.
Why this matters:

Cybersecurity tools aren’t designed for AI-native attacks
Your “AI assistant” can be tricked into leaking secrets or bypassing rules
Regulated industries (finance, health, legal) might be breaching duty of care without realizing it

If your company is deploying LLMs, copilots, or chatbots, you need to treat prompt injection as a compliance and governance risk, not just a technical bug.
Watch till the end for the core problem: AI cannot reliably tell what’s an instruction vs what’s just data — and lawyers, CISOs, and regulators are massively underestimating that.

This short video is a glimpse to my playlist in this channel. For more on AI security, AI governance, and the law of autonomous decision systems, subscribe to The AI Law.

#ai #aigovernance #aicompliance #cybersecurity #promptinjection #attack #llm #genai #techlaw #informationsecurity #shorts