1. Home
  2. Containment Board
  3. Z.ai (Zhipu)

Containment Board · rank 9 of 11 · updated Nov 8, 2026

Z.ai (Zhipu) safety report card

Z.ai, formerly Zhipu, is a Chinese SI developer that makes the GLM models. On the Containment Board it gets D overall (GPA 0.86), rank 9 of 11. Z.ai, formerly Zhipu AI, signed the Seoul commitments and answered FLI's survey, and FLI rates it more transparent than its Chinese peers, but it has no published framework or whistleblower policy. It told FLI it commissioned no outside pre-release testing, and US CAISSI later called GLM-5.3 the most cyber-capable open-weight model released to date.

Overall
D
GPA
0.86
Rank
9 of 11
Sources
4

Five grades

Each grade, why, and the sources

The same rubric applies to every lab. Grades use public documents only, current to November 8, 2026.

Safety framework

DHow it is graded

Why D

No published frontier safety framework was found; Z.ai signed both the Seoul Frontier AI Safety Commitments, which ask signatories to set out thresholds, and China's CAICT safety commitments.

D on this criterion means: No framework of its own; the lab has only signed voluntary pledges such as the Seoul Frontier AI Safety Commitments, the EU General-Purpose AI Code of Practice or China's CAICT commitments.

Sources

  1. AI Safety Index Summer 2026, full report: Safety Frameworks domain Future of Life Institute ·
  2. AI Safety Index Summer 2026, full report: Voluntary Commitment indicators (Seoul commitments, EU Code of Practice) Future of Life Institute ·
  3. Frontier AI Safety Commitments, AI Seoul Summit 2024 UK Department for Science, Innovation and Technology ·
  4. Frontier AI Safety Policies (index of published company frameworks) METR ·

Outside testing

DHow it is graded

Why D

In its FLI survey answer Z.ai said no outside pre-release testing was commissioned, although its Seoul pledge asks for risk assessments to consider outside evaluations. US CAISSI assessed GLM-5.2 and GLM-5.3 after release.

D on this criterion means: A commitment to outside testing exists (own framework or signed pledge), but no evidence of pre-release outside testing of the flagship.

Sources

  1. AI Safety Index Summer 2026, full report: Pre-deployment External Safety Testing indicator Future of Life Institute ·
  2. Frontier AI Safety Commitments, AI Seoul Summit 2024 UK Department for Science, Innovation and Technology ·
  3. CAISI's Assessment of Z.ai's GLM-5.3 Cyber Capabilities NIST ·

Shutdown and pause

DHow it is graded

Why D

Z.ai has no own pause commitment beyond its Seoul pledge, which has no published thresholds. In its FLI survey answer it described internal deployment safeguards such as logging all inputs and outputs, live monitoring and automated editing of suspicious outputs.

D on this criterion means: Discretionary language only ('may shut down', 'as appropriate'), or a signed pledge not to develop or deploy with no published thresholds that would trigger it.

Sources

  1. Frontier AI Safety Commitments, AI Seoul Summit 2024 UK Department for Science, Innovation and Technology ·
  2. AI Safety Index Summer 2026, full report: Internal Monitoring and Control Interventions indicator Future of Life Institute ·

Incident disclosure

D+How it is graded

Why D+

Z.ai answered FLI's survey and described incident response infrastructure, but public incident reporting rests on Chinese legal duties and no incident or misuse reports were found.

D on this criterion means: Minimal safety documentation; incident reporting only where the law requires it.

Sources

  1. AI Safety Index Summer 2026, full report: Serious Incident Reporting and Government Notifications indicator Future of Life Institute ·
  2. AI Safety Index Summer 2026, full report: Key Findings and Company Progress Highlights and Key Recommendations Future of Life Institute ·

At a glance

Models, facts and its own rules

The flagship models the board looked at, the facts that shaped the grades and the safety policy Z.ai (Zhipu) has published for itself.

Flagship models

GLM-5.3

Notable

Facts behind the grades

  • US CAISSI called GLM-5.3, released Aug 14, 2026, the most cyber-capable open-weight model released to date, about four months behind the US frontier.

Its own rules

Published safety policy

The Control Map lists no frontier safety framework of its own for Z.ai (Zhipu). Its safety framework grade above explains what the board found instead.

The Rogue Log

Incidents involving Z.ai (Zhipu) systems

Cases in the Rogue Incident Database that involve Z.ai (Zhipu) systems.

The Rogue Incident Database has no documented case involving Z.ai (Zhipu) systems yet. That can mean a clean record or little disclosure; the incident disclosure grade above says which is more likely.

Browse all 59 incidents

The Control Map

SI safety rules where it is based

Z.ai (Zhipu) is based in China. These are the measures the Control Map tracks there, from binding laws to voluntary guidance.

Cite this report card

"Z.ai (Zhipu) safety report card." The Containment Board, enterrsi.si, November 8, 2026. Overall D (GPA 0.86), rank 9 of 11. https://enterrsi.si/containment-board/zai/

Free to cite under CC BY 4.0 with a link to enterrsi.si.

Every lab on the board

Anthropic B+OpenAI BGoogle DeepMind B-Microsoft C+Meta C+Amazon CxAI D+Mistral DAlibaba (Qwen) D-DeepSeek F

Enter RSI

See the whole board

11 labs, five criteria, one public method. Sort them, compare them and download the data.

Open the Containment Board Incident Database