Advancing Cybersecurity with AI and Machine Learning: Opportunities and Challenges
summary
Panelists give concrete claims and tradeoffs: ML improves log and threat-feed correlation so SIEM/EDR catch more anomalies but never reaches 100% — quoted numbers include 95% quorum tolerance, “1%” residual risk and survey figures like 91% gen use vs 65% lack of understanding. Examples include a vulnerability platform kikimora IO, a fraud-blocking CVV bug, RAG on top of LLMs, and attacker malware that mutates; humans-in-loop and regulation (EU AI Act) are proposed mitigations.
AI-written summary · about
chapters (13)
- 0:00AI in cybersecurity intro
- 1:05Security definitions panel
- 6:04AI for log analysis
- 11:43AI model safety
- 13:11AI in finance
- 17:03RAG for security
- 20:33AI reliability in security
- 23:46Human oversight in AI security
- 26:35Developer education on AI risks
- 29:01Generative AI security
- 31:09Ethical AI in security
- 37:28Q&A on AI threats
- 44:12Closing remarks
related talks
44mAdvancing Cybersecurity with AI and Machine Learning: Opportunities and Challenges
32mAI/Machine Learning: What is Actually Working in Cybersecurity?
49mCXO Panel: Cybersecurity: Is Machine Learning The Answer?
45mHardening AI/ML Systems - The Next Frontier of Cybersecurity
11mAI-Powered Cybersecurity in the Age of Advanced Threats
16mMachine Learning--Cybersecurity Boon or Boondoggle
25mHuman-machine intelligence in cybersecurity
44mAI Meets Cybersecurity: Boosting Speed Without Losing Control