<!-- mobian-agent-page publisher="dailydev" canonical="https://daily.dev/posts/z-ai-and-concordia-ai-propose-six-stages-for-managing-open-weight-ai-risk-ldtwtnxmx" -->

---
title: Z.ai and Concordia AI propose six stages for managing...
description: Chinese AI company Z.ai and Beijing-based Concordia AI have published a framework for managing risks of open-weight AI models, whose parameters anyone can...
canonical: https://daily.dev/posts/z-ai-and-concordia-ai-propose-six-stages-for-managing-open-weight-ai-risk-ldtwtnxmx
twitter:card: summary_large_image
twitter:site: @dailydotdev
og:type: website
og:site_name: daily.dev
og:title: Z.ai and Concordia AI propose six stages for managing open-weight AI risk | daily.dev
og:description: Chinese AI company Z.ai and Beijing-based Concordia AI have published a framework for managing risks of open-weight AI models, whose parameters anyone can...
og:url: https://daily.dev/posts/z-ai-and-concordia-ai-propose-six-stages-for-managing-open-weight-ai-risk-ldtwtnxmx
og:image: https://api.daily.dev/og/posts/LdtwTNxmX.png
og:image:alt: Z.ai and Concordia AI propose six stages for managing open-weight AI risk
og:image:width: 1200
og:image:height: 630
og:locale: en
---

> ## Documentation Index
> Fetch the complete documentation index at: https://daily.dev/llms.txt
> Use this file to discover all available pages before exploring further.

# Z.ai and Concordia AI propose six stages for managing open-weight AI risk

**[The Next Web](https://daily.dev/sources/tnw)** · 2 min read · 0 upvotes · 0 comments

## Summary

Chinese AI company Z.ai and Beijing-based Concordia AI have published a framework for managing risks of open-weight AI models, whose parameters anyone can download, modify and run. The framework addresses three core problems: releases are irreversible, safety training can be stripped out with minimal fine-tuning, and developers lose visibility into usage after release. It proposes six stages: risk identification, risk thresholds (assessed across where the model runs, who might misuse it, what it enables, and societal absorption capacity), risk analysis, risk evaluation (sorting models into green/yellow/red release zones), risk mitigation, and risk governance. Z.ai's GLM-5.3 release, where vetted partners tested the model before full weights were published, is cited as an example of staged release in practice. The report follows Shanghai AI Lab and Concordia AI's Frontier AI Risk Management Framework 2.0 from July.

## Full article

daily.dev links to this article rather than hosting it. Read it at the original source: <https://thenextweb.com/news/z-ai-concordia-open-weight-ai-risk-framework-six-stages>

## Questions this post answers

### What is the staged release approach Z.ai used for GLM-5.3?

Vetted security partners tested GLM-5.3 first, and the full model weights were only published after safety evaluations were completed. This staged release approach is cited as a practical example within a six-stage open-weight AI risk management framework published jointly by Z.ai and Concordia AI, which recommends staged release instead of a binary open-or-closed decision.

_Anyone weighing open-weight model releases can follow ongoing AI safety framework coverage on daily.dev._

### Why is it hard to make open-weight AI models safe after release?

Once model weights are publicly released, the process is essentially irreversible, and safety training can be undone with just a small number of harmful fine-tuning examples. Developers also lose the ability to monitor how downstream users actually run or modify the model, which is why risk mitigation measures need to work before release or continue functioning after weights are out.

_Developers tracking open-weight model safety debates can follow the discussion on daily.dev._

## Similar posts on daily.dev

- [Open-weight AI caught the frontier on capability. On safety, it didn’t](https://daily.dev/posts/open-weight-ai-caught-the-frontier-on-capability-on-safety-it-didn-t-onyvj8soa) · The Next Web · 0 upvotes · 0 comments
- [The Cyber Risk Discourse is Broken](https://daily.dev/posts/the-cyber-risk-discourse-is-broken-w1ckoqv8l) · Interconnects · 0 upvotes · 0 comments
- [Microsoft, Nvidia push provenance checks as open-weight AI adoption accelerates](https://daily.dev/posts/microsoft-nvidia-push-provenance-checks-as-open-weight-ai-adoption-accelerates-kten8x7q1) · CSO Online · 3 upvotes · 0 comments
- [Weekly Dose \#9 - AI Is Becoming an Access-Control Problem](https://daily.dev/posts/weekly-dose-9---ai-is-becoming-an-access-control-problem-odg9xd3us) · Machine Learning Pills · 2 upvotes · 0 comments

---

Tags: [#llm](https://daily.dev/tags/llm), [#ai-safety](https://daily.dev/tags/ai-safety), [#ai-governance](https://daily.dev/tags/ai-governance)

[View this post on daily.dev](https://daily.dev/posts/z-ai-and-concordia-ai-propose-six-stages-for-managing-open-weight-ai-risk-ldtwtnxmx)

```json
{"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]}
{"@context":"https://schema.org","@type":"TechArticle","headline":"Z.ai and Concordia AI propose six stages for managing open-weight AI risk","url":"https://daily.dev/posts/z-ai-and-concordia-ai-propose-six-stages-for-managing-open-weight-ai-risk-ldtwtnxmx","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/z-ai-and-concordia-ai-propose-six-stages-for-managing-open-weight-ai-risk-ldtwtnxmx"},"datePublished":"2026-09-28T17:01:06.436Z","dateModified":"2026-09-28T17:01:26.608Z","description":"Chinese AI company Z.ai and Beijing-based Concordia AI have published a framework for managing risks of open-weight AI models, whose parameters anyone can...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/a2a7db3d212855d9d870573babdd2a02?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/a2a7db3d212855d9d870573babdd2a02?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"The Next Web","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"The Next Web","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/tnw","url":"https://daily.dev/sources/tnw"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/z-ai-and-concordia-ai-propose-six-stages-for-managing-open-weight-ai-risk-ldtwtnxmx","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"llm,ai-safety,ai-governance","timeRequired":"PT2M"}
{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"The Next Web","item":"https://daily.dev/sources/tnw"},{"@type":"ListItem","position":3,"name":"Z.ai and Concordia AI propose six stages for managing open-weight AI risk"}]}
{"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/z-ai-and-concordia-ai-propose-six-stages-for-managing-open-weight-ai-risk-ldtwtnxmx#faq","mainEntity":[{"@type":"Question","name":"What is the staged release approach Z.ai used for GLM-5.3?","acceptedAnswer":{"@type":"Answer","text":"Vetted security partners tested GLM-5.3 first, and the full model weights were only published after safety evaluations were completed. This staged release approach is cited as a practical example within a six-stage open-weight AI risk management framework published jointly by Z.ai and Concordia AI, which recommends staged release instead of a binary open-or-closed decision. Anyone weighing open-weight model releases can follow ongoing AI safety framework coverage on daily.dev."}},{"@type":"Question","name":"Why is it hard to make open-weight AI models safe after release?","acceptedAnswer":{"@type":"Answer","text":"Once model weights are publicly released, the process is essentially irreversible, and safety training can be undone with just a small number of harmful fine-tuning examples. Developers also lose the ability to monitor how downstream users actually run or modify the model, which is why risk mitigation measures need to work before release or continue functioning after weights are out. Developers tracking open-weight model safety debates can follow the discussion on daily.dev."}}]}
```

