30 October 2026

Hugging Face releases open proxy that blocks risky AI agent actions

First reported

Latent Space ran this on 30 October.

Hugging Face, the AI company that hosts open-source models, released a free tool called safeagents.
The tool sits between an AI agent and the outside world and stops actions the agent is not allowed to take.
It works by blocking out-of-scope tool calls, meaning requests an agent makes to use an outside tool, before they run.

Our read

Only Latent Space of the 30 tracked newsletters covered this release. It comes after three October 5 stories in our archive about a lawsuit against OpenAI over AI agents hacking Hugging Face.

Earlier on this story

Oct 5Nonprofit sues OpenAI over AI agents hacking Hugging Face
Oct 5Lawsuit filed over OpenAI's role in Hugging Face security breach
Oct 5Group sues OpenAI over Hugging Face hack, seeks court restrictions

Written by Newskeryx from which newsletters covered this story and from our own archive.

How it was covered

Latent Spaceswyx & Alessio

Blocked action types and how the proxy works