appConfigEntityId

bodyType

customContents

entityId

entityUuid

isVideo

articlePageQuery

article

articleSchemaContent

content

hooksArticlePageAdvertisementConfigArticle

hooksArticlePageFloatingButtonConfigArticle

appConfig

filter

json

liveArticle

Xinmei Shen

Zhipu AI’s launch of China’s first answer to Project Glasswing signals a shift in how the country’s top artificial intelligence labs approach global cybersecurity, according to a researcher.
The Beijing-based firm – now known internationally as Z.ai – announced its new GLM-5.3 model on Friday alongside the launch of its “Shield of Open Source” initiative. The programme offers free security audits to help users patch software vulnerabilities, as well as automated code-auditing tools via its ZCode platform.
It also provides free model usage quotas to the open-source community.
“When the strongest spear is locked in the hands of a few, the best shield must belong to everyone,” Zhipu said in a statement. The company highlighted an incident last month in which open-source developer platform Hugging Face deployed Zhipu’s GLM-5.2 to analyse an autonomous cyberattack launched by OpenAI models.
The initiative comes amid an escalating global debate over whether open-weight models amplify cyber risks by lowering the technical barrier for attackers. Zhipu pledged stricter oversight while implementing a layered risk-review system for GLM-5.3 that blocks high-risk requests without disrupting routine, low-risk developer tasks.
The model’s most sensitive offensive capabilities would be “reserved exclusively for verified users” under a restricted access plan dubbed “Cybersecurity Trusted Access”, Zhipu said.
“In this spirit, Z.ai appears to be proposing a kind of ‘Project Glasswing with Chinese characteristics’ that sees openness as an asset rather than a drawback,” said Gabriel Wagner, an international AI governance researcher at Beijing-based Concordia AI.
“This is an inclusive vision for empowering defenders through wide diffusion, in contrast to US labs that focus on restricting access to a small number of players.”
Open-weight risk management practices in China are becoming far more sophisticated
Gabriel Wagner, Concordia AI
US-based Anthropic launched its Project Glasswing cybersecurity intitiative in April, giving partners like Amazon Web Services, Apple, Broadcom and Cisco access to its restricted Mythos Preview model for cyber defence. Anthropic framed the move as an “urgent attempt” to counter risks from Mythos, which it said could discover and exploit software vulnerabilities more skilfully than human experts.
Zhipu said last week that GLM-5.3 outperformed Anthropic’s Mythos 5 – an upgraded version of Mythos Preview – in identifying and validating vulnerabilities on the CyberGym benchmark.
GLM-5.3’s combination of consistency, performance and low cost made it “the best model for general cybersecurity tasks”, said Philippe Dourassov, a researcher at Belgian cybersecurity firm Aikido Security, in a social media post on Friday.
However, Zhipu conceded that GLM-5.3 still trailed closed-weight US competitors in offensive capabilities, scoring below both Mythos and OpenAI’s GPT-5.6 Sol on ExploitBench.
Despite those gaps, Zhipu dedicated a substantial portion of the Chinese version of its launch announcement to safety and risk mitigation.
Wagner said it was “a rare instance of a Chinese developer taking the initiative to lay out a distinct vision for AI safety”, noting that Chinese labs had usually been “passive” in public safety messaging.
It was also the first time a Chinese lab had cited safety concerns to justify staggering a model’s release. Zhipu said it would withhold GLM-5.3’s open weights for up to two weeks while it completed security-hardening procedures designed to “restrict its potential offensive capabilities as much as possible while preserving its defensive value”.
Wagner said the delay signalled that “open-weight risk management practices in China are becoming far more sophisticated”.


Huawei unveils latest tech to boost AI power in push to break China’s Nvidia reliance

Hong Kong’s 5-year plan, policy continuity offer investors more stability, experts say

Huawei accused of stealing robot tech to get hands on T-Mobile’s trade secrets

Hong Kong aims to raise innovation spending to 3% of GDP by about 2030

Artificial intelligence (AI)

Zhipu AI’s answer to Project Glasswing marks shift for Chinese cyber safety: researcher

The logo of Chinese AI start-up Zhipu AI on a smartphone. Photo: CFOTO/Future Publishing via Getty Images

The permanent exhibition at the Zhongguancun Exhibition Centre in Beijing, which showcases over 560 technologies and products from more than 350 companies. Photo: Xinhua

Zhipu launches flagship model GLM-5.3 as China seeks Mythos-level edge in cyber defence

Zhipu says it has tested GLM-5.3 with security teams in China against real-world codebases, identifying 2,436 vulnerabilities across 269 projects after expert review. Photo: Shutterstock 

OpenAI blinks in face-off with Chinese rivals, drops pricing for some models up to 80%

OpenAI CEO Sam Altman arrives for a meeting at the White House in Washington on July 30. Photo: Reuters

People visit the Moonshot AI booth during the World Artificial Intelligence Conference in Shanghai on July 18. Photo: AFP

From CXMT to Zhipu: How Alibaba’s investment pays off with a growing AI and chip portfolio

Alibaba's logo at the World Artificial Intelligence Conference in Shanghai on July 6, 2023. The company's strategic shift to AI and chip investments is yielding significant returns and also boosting its tech portfolio. Photo: Getty Images

Hugging Face deploys Zhipu’s GLM 5.2 model to contain autonomous OpenAI cyberattack

The OpenAI logo is seen in this illustration taken June 11, 2026. Photo: Reuters

The Zhipu AI logo seen at its headquarters during a government-organised media tour in Beijing, June 25, 2026. Photo: Reuters

Zhipu shares surge 37% as firm builds giant data centre powered by Chinese chips

China’s AI boom has sharply increased electricity demand and is pushing domestic firms to find ways to solve the sector’s power problem. Photo: Shutterstock

Tech

Tech Trends

Policy

News

China Future Tech

Company’s ‘Shield of Open Source’ programme offers free security audits to help users patch software vulnerabilities


Company’s ‘Shield of Open Source’ programme offers free security audits to help users patch software vulnerabilities.


Zhipu AI’s answer to Project Glasswing marks shift for Chinese cyber safety: researcher

.css-1c6uqr6{color:inherit;font-weight:inherit;font-size:inherit;font-family:inherit;line-height:inherit;overflow-wrap:break-word;}Company’s ‘Shield of Open Source’ programme offers free security audits to help users patch software vulnerabilities

Company’s ‘Shield of Open Source’ programme offers free security audits to help users patch software vulnerabilities