CC13: Top 10 Unexpected AI Systems Hacks that we did not see coming! скачать в хорошем качестве

CC13: Top 10 Unexpected AI Systems Hacks that we did not see coming!

From invisibility cloaks that fool thermal cameras to deep‑fake voices that empty bank accounts—see the ten strangest AI hacks shaking every industry. AI dorks are the new Google dorks. Drawing on 50 + real incidents from the past three years, Gopi and Aruneesh rank the Top 10 unexpected attacks and map them to business, consumer, and critical‑infrastructure impact. You’ll learn: Org‑level shocks Cloaking wearables that blind vision‑AI door controls Voice jailbreaks that bypass audio guardrails Poisoned foundation models hiding in public repos Consumer nightmares Celebrity deep‑fake romance scams ($800 K loss) Augmented‑reality “look‑and‑steal” glasses that profile victims in seconds Voice‑clone account takeovers at banks without MFA Macro‑scale threats Model‑extraction leaks slashing $600 B in market cap AI‑driven flash‑crash scenarios flagged by regulators Each hack is scored for likelihood, impact, and prevention difficulty and paired with practical mitigations—from stricter API‑key controls to circuit breakers for algorithmic trading. Chapters 00:00 Welcome & talk outline 00:29 Speaker intros – Gopi & Arun 08:38 AI ecosystem cheat‑sheet (producers, deployers, bad actors) 10:12 Known vs unexpected AI hacks 14:48 Org Hack #1 – Invisibility cloak vs thermal cameras 17:56 Org Hack #2 – Voice jailbreak prompt injection 20:46 Org Hack #3 – Supply‑chain model swap on public hubs 22:48 Org Hack #4 – Data‑poisoning fraud models (finance) 25:51 Consumer Hack #1 – Celebrity deep‑fake love scam 28:19 Consumer Hack #2 – AR glasses instant doxxing 30:46 Consumer Hack #3 – Voice‑clone bank TFA bypass 34:35 Infra Hack #1 – Flash‑crash risk & SEC warning 36:57 Infra Hack #2 – Model‑extraction crash (DeepSeek case) 40:12 Mitigation playbook – guardrails, audits, circuit breakers 41:50 Key takeaways & recommended frameworks (MITRE ATLAS, NIST AI RMF) 46:00 Q&A and resource links #aihacks #AdversarialAI #CactusCon #ModelSecurity #DeepFakeFraud