Post flags publicly exposed large AI models, including Kimi 1T and DeepSeek 765B

max_paperclips · x · 2026-07-27

This post points to a controversial security issue: AI models exposed on the public internet and left accessible to others without proper protection.

The quoted text specifically mentions very large models such as Kimi 1T and DeepSeek 765B, framing the situation as people freely using systems that were never meant to be public. While the post is informal, the underlying topic is a real AI security and privacy problem: exposed inference endpoints and the risks of leaving powerful models open to the internet.

Original post →

More from Safety

Safety channel →