Data Security in the AI Era, Don't Let AI Ruin your Presentation, and the Weekly News - Hadar Krasner - ESW #480
Summary
Adel Krasner of Bold Security discusses endpoint data protection, emphasizing the need for context-aware solutions to manage AI risks and insider threats. The hosts then critique the quality of AI-generated presentation slides, highlighting issues with semantic incoherence and the importance of human curation. Finally, the episode covers enterprise security news, including the "sawtooth problem" and the rise of data extortion groups that steal information without deploying ransomware.
Topics discussed
Episode intro and sponsor mentions
FinSec conference promotion
Interview intro: Bold Security and endpoint data
The importance of protecting enterprise data
Why traditional DLP fails in the AI era
Reframing the question: Identity and data on endpoint
Ransomware and data exfiltration vectors
Keeping honest users honest with AI tools
Shadow IT and unauthorized SaaS usage
Product philosophy: Problem plus solution
Founding story and CISO insights
Limitations of legacy DLP: Context and cost
Early tools and reducing false positives
Case study: Credit card data in browser caches
Role-based context and data lineage
Integrations and the burden of manual configuration
Automated discovery and self-mapping
Articulating data usage per business unit
Handling local AI agents and legacy tools
OCR and the ease of data theft from printers
Tracking intent and proactive protection
Workshop results: Zero hands raised
Reducing the customization tax
Balancing automation with flexibility
Customer success story: Self-service adoption
Interview wrap-up and call for questions
Sponsor: Travelers Cyber Threat Report
Sponsor: ThreatLocker
Topic: AI-generated slides and quality pitfalls
The struggle of making slides manually
Proposal for an 'Adrian-approved' slide skill
Slides vs. Talk tracks: The mismatch
The need for minimalist slides
Slides as code and HTML presentations
Numbering and ordering presentation materials
Example of good slide navigation cues
Effort levels: Informal vs. formal talks
Last-minute slide creation in airports
Performance under pressure
AI for structure vs. AI for content
AI vocabulary errors: 'Ascend' vs. 'Rise'
The importance of reading output aloud
Sponsor: Horizon 3 Node Zero
News segment intro and Vibe Check setup
Vibe Check: Shrinking exposure vs. fixing risk
Adrian's opinion: Fixing existing risk
The case for shrinking exposure
Vibe Check results and analysis
Preventative controls and forward-looking security
Is shrinking exposure an investable domain?
Tools vs. processes in exposure reduction
Personal anecdote: Unpatched software
Is the Vibe Check a topic segment?
Fastest growing category: Automated red teaming
VC funding and traction in red teaming
News: Ransomware negotiator charged in Florida
Insights from ransomware negotiators
News: Engineer ransoms own company
Recalling the Bart ransomware incident
News: Silent Ransom Group targets law firms
Reputation and repeat victims in extortion
Payment statistics for Silent Ransom Group
How the engineer was caught: Court documents
Weak password: 'frozen crew!'
Behavioral issues and credential cleanup
The 'Alter Ego' approach to insider threats
News: Shiny Hunters and FBI data breach
Accountability for unpatched vulnerabilities
Blame game: Interns vs. Third parties
Complexity of patching and change control
Reframing security failures in other industries
What Americans want from companies: Honesty
Living up to corporate security values
Microsoft's security track record
Cybersecurity in non-security conferences
Future of NPM: AI writing its own libraries
AI pathfinding and code consistency
Supply chain risks and library independence
Non-engineers building software companies
Essay: Passkeys for enterprise vs. consumer
Passkey usability issues and device sync
Account recovery and lockout concerns
Usability advantages of passkeys
Squirrel story: AI pulling images from minds
Outro and next week's preview
Listen ad-free on Castria