Use Generative AI to End Your Love/Hate Relationship with DLP
summary
Argues that DLP must be an approach not a single product, mapping controls to concrete Gen AI scenarios: DLP as an outcome for BYO public tools, embedded copilots, tuned models and rare custom training; recommends discovery/classification, data access governance, DLP for prompts/responses, and privacy-preserving tech (masking, tokenization, homomorphic encryption, confidential computing) plus vendor due‑diligence questions. Cites a July finding that 37% already update policies, warns watermarking is immature, and frames a 3–6 month roadmap to close gaps.
AI-written summary · about
chapters (14)
- 0:00DLP challenges and Gen AI solutions
- 3:56DLP as data security outcome
- 8:29Gen AI regulatory considerations
- 10:59Organizations and LLMs
- 13:51Data control in Gen AI workflows
- 15:16DLP policies for public AI tools
- 17:01AI copilots in enterprise tools
- 22:22Internal Gen AI applications
- 25:59Custom models and governance
- 28:13AI output authenticity
- 32:54Gen AI use policies
- 35:29Gen AI data control tech
- 40:10DLP implementation next steps
- 43:44Q&A on AI security
related talks
45mCombating Generative AI's Privacy Abuses
49mDLP: An Implementation Story
43mPanel: Privacy Design Patterns for AI Systems: Threats and Protections
18mAI Is Coming For Our Secrets
06mData Privacy Concerns and Solutions When using LLMs in Production
09mSecuring Generative AI Workloads
01h 06mDatabases and AI Training
49mThe Intersection of AI and Cybersecurity