惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

P
Privacy & Cybersecurity Law Blog
Engineering at Meta
Engineering at Meta
Forbes - Security
Forbes - Security
MongoDB | Blog
MongoDB | Blog
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
A
About on SuperTechFans
量子位
Exploit-DB.com RSS Feed
Exploit-DB.com RSS Feed
雷峰网
雷峰网
腾讯CDC
P
Proofpoint News Feed
S
Schneier on Security
S
Secure Thoughts
V
Visual Studio Blog
Help Net Security
Help Net Security
The Hacker News
The Hacker News
C
Cyber Attacks, Cyber Crime and Cyber Security
P
Privacy International News Feed
SecWiki News
SecWiki News
S
SegmentFault 最新的问题
T
Threatpost
小众软件
小众软件
MyScale Blog
MyScale Blog
F
Fortinet All Blogs
cs.CL updates on arXiv.org
cs.CL updates on arXiv.org
P
Proofpoint News Feed
T
Tailwind CSS Blog
I
Intezer
C
CERT Recently Published Vulnerability Notes
U
Unit 42
V
V2EX
Cyberwarzone
Cyberwarzone
Recorded Future
Recorded Future
O
OpenAI News
Project Zero
Project Zero
有赞技术团队
有赞技术团队
Google DeepMind News
Google DeepMind News
Last Week in AI
Last Week in AI
Hugging Face - Blog
Hugging Face - Blog
Know Your Adversary
Know Your Adversary
C
Cybersecurity and Infrastructure Security Agency CISA
Scott Helme
Scott Helme
V2EX - 技术
V2EX - 技术
博客园 - 叶小钗
S
Securelist
A
Arctic Wolf
The Cloudflare Blog
W
WeLiveSecurity
T
Threat Research - Cisco Blogs
博客园 - Franky

Mozilla.ai

From Evaluation to Guardrails: What We Brought to ACM FAccT 2026 Open Models are ready for agents. Their APIs are not. The Control Layer: Why the Next Era of AI Is About Infrastructure, Not Just Models Introducing Otari: The Open-Source LLM Control Plane Announcing transcribe.cpp Using Octonous as a Product Manager Image Classification Comes to encoderfile What is an LLM control plane? Use the Otari Gateway with OpenCode Otari: Own Your AI Stack | AI Gateway & Hosted Platform AI Got Expensive. Now What? | Mozilla.ai cq exchange: Agents without Borders The Interface Is No Longer the Product VIBE✓: First Defense for cq (Stack Overflow for Agents) Octonous Open Beta: What We've Learned and Where We're Going Sovereign AI: Control, Choice, and Beyond Geopolitics Encoderfile’s New Format: Why a “Dull” Design Wins The Real Challenge Behind Small Trade Businesses Hardening Your LLM Dependency Supply Chain cq: Stack Overflow for Agents cq: Stack Overflow for Agents llamafile Reloaded: What’s New in v0.10.0 When Shipping Software Becomes Too Easy Mozilla.ai Joins Flower Hub as Launch Partner Owning Code in the Age of AI The Star Chamber: Multi-LLM Consensus for Code Quality
Using Octonous as an AI Safety Engineer
Daniel Nissani · 2026-07-09 · via Mozilla.ai
Case Study

Discover how Octonous helps our team at Mozilla.ai cut through daily busywork, from automatically monitoring our libraries to staying on top of the latest developments in AI safety.

Using Octonous as an AI Safety Engineer

This is the second article in a series on how the Octonous team uses Octonous in their day-to-day work. Each post comes from someone on the team, writing about their own role.

The role of an AI Safety Engineer is multifaceted. Not only am I the core maintainer of any-guardrail, an open-source package that allows you to switch between guardrails seamlessly with an output standard that limits code edits, but I also conduct evaluations, security scans, and policy development. Such a broad role requires tools that make it easy to access many different integrations. Octonous is my go-to platform when I want my tools to talk to each other. In this blog post, we’ll go over three ways I use Octonous to help me do my work.

Creating Content Moderation Policies

cq is a centralized platform that we built that allows users to push knowledge units (KUs), resolution paths to errors their coding agents encounter. In the future, we plan to allow users to nominate their knowledge units to be used by other people’s coding agents. The risk surface of a platform like cq is vast, as we describe in our CAIS workshop paper, including, but not limited to, prompt injection attacks, DDoS attacks, and identity spoofing.

One way we plan to mitigate these risks is by having content moderation reviewers for nominated KUs. Octonous became an assistant for me, helping me aggregate and synthesize our documentation, allowing me to focus on refining my ideas for the types of content we want moderators to flag, as well as how to operationalize such content.

Getting Automated Updates on any-guardrail

One of the most frustrating things I’ve experienced is having a new issue or PR put up on any-guardrail, and I forget to triage it in a timely manner. This has resulted in some clunky conversations that could have gone smoother if I had some kind of notification of new issues and PRs that are put on any-guardrail. Luckily, I was able to create a daily message with Octonous that sends me an email about any changes that have been made any-guardrail. Now, I have a daily roundup, making it near impossible for me to miss out on proposed changes to the package.

Weekly AI Safety Papers

Part of my job is keeping up with the latest trends in AI safety. With such a broad role, it is hard to carve out time to find new papers, blog posts, etc., that will help spark new ideas and keep me up to speed. That’s why I used Octonous to create a weekly AI Safety round up. Every Friday morning, Octonous sends me a summary of the week in research with citations, so I can read the original source. This has saved me hours and lets me keep up with the field as I do the other components of my job.

Just Scratching the Surface

These are the components of my job that Octonous has impacted the most. But I believe I’m just scratching the surface. Octonous offers an easy user experience that allows you to link the applications you use most, unlocking workflows that weren’t possible before. I can’t wait to get more creative and figure out what else Octonous can help me do. Give it a try!

You can try Octonous at octonous.com. New users receive 1,000 free credits when they sign up, with more available through onboarding.