A secure LLM system cannot be given write access to any important endpoints. That’s the basic first principle of AI security at the moment.

I have a relatively high value Instagram username that I’m holding onto for dear life against a barrage of attackers trying to get Meta to give it to them. For Meta this is clearly not a priority and I could be losing this part of my online presence at any moment.

https://thecybersecguru.com/news/instagram-meta-ai-vulnerability-account-recovery-exploit/