Cloud Security Podcast Adapting to New Threats, Copilot Risks & The Future of Data (Feat. Matthew Radolec, Varonis)
16 snips
Jun 3, 2025 
Matthew Radolec leads systems engineering and incident response at Varonis, passionately dedicated to data protection. He delves into the surge of AI-driven phishing and the risks posed by tools like Copilot. The conversation highlights the alarming statistic that 86% of attacks stem from credential misuse and discusses the rising threat of AI-powered social engineering. Radolec emphasizes the critical need for robust access management and proactive measures in a landscape increasingly dominated by AI, ensuring data integrity amidst evolving challenges.
AI Snips
Chapters
Transcript
Episode notes
Measure AI Success and Failure
- Measure AI success by comparing outcomes against established metrics to ensure gains.
- Monitor for increased false positives or negatives to avoid hurting business with AI decisions.
Data Protection Techniques Resurge
- Masking, tokenization, and encryption will become crucial to protect sensitive data feeding AI models.
- Businesses must balance moving fast with limiting sensitive data exposure in AI training.
AI Model Security Responsibility
- AI model security responsibility lies mostly with developers, not consumers.
- Consumers cannot fix flaws in models like logic and reasoning errors on their own.