Скачать или смотреть Running Local Models Isn't More Secure, Unless You Do It Right—Steve Wilson Live with Tim O'Reilly

Watch the entire conversation: https://learning.oreilly.com/videos/b...

Running local models may seem like a more secure choice, but you still need to be concerned with the security of the supply chain. What’s the provenance of the model you’re planning to use, and who may have already comprised it? As Exabeam chief AI and product officer Steve Wilson warns, you should always treat your model “as something in between a confused deputy and an enemy sleeper agent.” Don’t let that put you off from trying out models—just be aware of the risks.

And if you’d like to do a little hands-on experimenting, Steve’s created Steve's Chat Playground: an open source project that lets you test chat models, personalities, guardrails, and user interfaces locally. Check it out on GitHub: https://github.com/virtualsteve-star/...

Follow O'Reilly on:
LinkedIn:   / oreilly  
Facebook:   / oreilly  
Instagram:   / oreillymedia  
BlueSky: https://bsky.app/profile/oreilly.bsky...