[gok]

> While they provide data privacy, our research shows their weaker reasoning and alignment capabilities make them easier targets for sabotage.

If you are using any LLM's reasoning ability as a security boundary, something is deeply, deeply wrong.

[liqilin1567]

This reminds me of stalwart's spam filter feature claim: "LLM-driven spam filtering and message analysis." :D

https://github.com/stalwartlabs/stalwart