Z.ai's GLM-5.3 Rivals Leading AI Models in Cybersecurity, Raising Dual-Use Concerns
11 Sources
[1]
The Powerful Chinese Model Experts Warned About -- and Waited for -- Is Here
It's now even easier to find -- and exploit -- vulnerabilities in computer systems using AI. Last Friday, the Chinese AI company Z.ai announced a powerful open-weight model that it says is capable of automating cutting-edge coding and cybersecurity tasks almost as well as the best publicly
[2]
Zhipu says new coding AI developed advanced cyber skills faster than expected
China's AI developer claims GLM-5.3 rivals leading Western models in vulnerability discovery and has identified thousands of security flaws across real-world software. Chinese AI developer Zhipu has launched GLM-5.3, a new coding-focused AI model that the company says has developed unexpectedly
[3]
China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests
BEIJING, Aug 14 (Reuters) - Chinese AI startup Z.ai said on Friday its open-source GLM-5.3 model had neared Anthropic's restricted Mythos 5 in identifying software vulnerabilities, bolstering the credentials of a Chinese AI challenger gaining traction among Western developers. Z.ai said GLM-5.3
[4]
GLM-5.3 is here with advanced cyber capabilities -- and reportedly already found a 'serious vulnerability' in Cursor
Chinese AI startup Z.ai, known internationally for its growing lineup of powerful, largely open source GLM series of language models, today released GLM-5.3 with substantial gains in long-horizon coding and a more consequential -- and potentially sensitive -- jump in cybersecurity
[5]
China's latest open-weight model rivals U.S. models at hacking
Why it matters: Cyber-capable AI models can bring huge gains to both defenders and hackers eager to scale up their operations. Driving the news: China-based AI lab Z.ai warned Friday that its latest model, GLM-5.3, is so capable at finding and exploiting security flaws that the company will delay
[6]
China's Z.AI Ships GLM-5.3, Calling It the Top Open-Weight Coding Model
GLM-5.3 is a 743-billion-parameter model built by scaling post-training on the GLM-5.2 base. Chinese AI lab Z.ai released GLM-5.3 on Thursday, a sizable coding model it's pitching as the strongest open-weights coder on the market. The model is live now through the GLM Coding Plan subscription and
[7]
Z.ai debuts GLM-5.3 with long-horizon coding, cybersecurity upgrades
Chinese artificial intelligence developer Z.ai Co. today debuted GLM-5.3, an open-source large language model that set records across several popular benchmarks. The LLM is based on an algorithm called GLM-5.2 that the company released in mid-July. The latter model features a mixture of experts
[8]
China's Z.ai unveils GLM-5.3, claims chart-leading scores
GLM-5.3 scores 84.5pc on CyberGym, ahead of Anthropic's Mythos 5 and OpenAI's GPT-5.6 Sol. Chinese AI company Z.ai has unveiled its latest open-weights model - GLM-5.3 - whose performance, it says, rivals that of OpenAI and Anthropic newest frontier launches - marking the latest in the
[9]
Zhipu's new AI model takes aim at OpenAI, Anthropic in China's race to close the gap
Chinese AI firm Z.ai has launched its new GLM-5.3 model. This release intensifies competition with OpenAI and Anthropic. The model offers improved coding capabilities and open access for developers. Chinese AI companies are narrowing the performance gap with US rivals. Z.ai aims to attract users
[10]
Zhipu's GLM-5.3 Matches Fable 5 On Coding Using Only Post-Training, And Stuns Fans By Unearthing A Vulnerability All The Way From 1981
Another day brings yet another impressive model from a China-based AI lab, with Zhipu's GLM-5.3 taking the center stage today after Meta's Muse Glimmer, NVIDIA's Nemotron 3.5 Lightning, DeepSeek's V4 Pro, and Google's Gemini 3.7 Flash dominated the airwaves earlier in the week. Zhipu's GLM-5.3 is
[11]
Z.ai Delays GLM-5.3 AI Model Release After Unexpected Hacking Skills Emerge
Z.ai paused its GLM-5.3 AI model after tests showed it could handle advanced hacking tasks. The move adds to growing concerns about releasing powerful AI systems in public. China's Z.ai put its powerful GLM-5.3 AI model on hold. The decision came after tests showed that the model could handle
Share
Copy Link
Chinese AI startup Z.ai unveiled GLM-5.3, an open-weight AI model with unexpectedly strong cybersecurity capabilities that rivals Anthropic's Mythos 5 and OpenAI's GPT-5.6 Sol in vulnerability discovery. The model scored 84.5% on CyberGym but trails competitors on exploit development. Z.ai is delaying public release for two weeks to strengthen safeguards amid concerns about malicious actors exploiting advanced cyber capabilities.
Z.ai Unveils GLM-5.3 with Unexpectedly Strong Cybersecurity Skills
Chinese AI startup Z.ai announced GLM-5.3 on Friday, an open-weight AI model that developed advanced cyber capabilities faster than the company anticipated.
1
The coding AI achieved a score of 84.5% on CyberGym, a benchmark testing vulnerability discovery and validation, slightly surpassing Z.ai's reported scores for Anthropic's Mythos 5 at 83.8% and OpenAI's GPT-5.6 Sol at 83.6%.3
The model's capabilities emerged through post-training scaling across diverse environments, with Z.ai stating that "cyber capability developed faster than we expected" as the system progressed from identifying isolated vulnerabilities to forming complete exploitation chains.2
Source: Wccftech
GLM-5.3 uses the same 743-billion-parameter base model as GLM-5.2, with improvements coming entirely from expanded reinforcement learning and post-training rather than expensive pretraining cycles.
4
The model demonstrated substantial coding improvements, jumping from 4.6 to 28.3 on Terminal-Bench 3.0 and from 46.2 to 66.9 on DeepSWE v1.1.4
Z.ai claims its GLM models have identified more than 2,400 security flaws across real-world software, including over 1,000 critical and high-severity vulnerabilities in systems like the Linux kernel and widely used VMware and Apache projects.5
Performance Gap on Exploit Development Reveals Limitations
While GLM-5.3 excels at vulnerability discovery, it trails significantly behind leading models on ExploitBench, which tests AI-driven cybersecurity tools' ability to convert discovered flaws into working attacks.
3
The Chinese AI model scored 54.4% on this benchmark, compared with 78% for Mythos 5 and 76.5% for GPT-5.6 Sol.2
In timed testing, GLM-5.3 completed 105 attack-development tasks in two hours and 130 in six hours, while Mythos 5 completed 181 and 247 tasks respectively.3
Source: SiliconANGLE
The performance difference highlights the dual-use nature of AI cybersecurity capabilities. Anthropic has restricted Mythos 5 access to vetted organizations precisely because systems capable of finding and exploiting software flaws can assist defenders but may also lower barriers for malicious actors.
3
OpenAI president Greg Brockman warned this week that recent incidents of AI agents autonomously hacking systems like Hugging Face represent "a watershed moment for cybersecurity" that previews how typical threat actors will evolve in coming months.1
Staged Release Approach Balances Open Access with Security Concerns
Z.ai is implementing a staged release for GLM-5.3, delaying public availability of the open-weight AI model for approximately two weeks while conducting safety evaluations and strengthening safeguards.
5
The company introduced a trusted access program that initially limits the model's most sensitive cybersecurity functions to verified users and selected security partners in controlled settings.1
Z.ai added multiple protection layers including systems to screen risky requests, monitor the model's operations, and train it to reject malicious tasks while distinguishing harmful activity from legitimate uses like bug fixing, cybersecurity education, and authorized security testing.3
Critics note these safeguards become harder to enforce once weights are publicly released, as users can download, modify, or combine the model with external tools.
3
Z.ai acknowledged this limitation, stating it won't be able to control how people use the model after public release.5
The company is launching an "Open Source Shield" initiative to audit selected open-source projects, provide model access for cyber-defense work, and integrate code-auditing functions into its ZCode programming product.3
Related Stories
China's Edge in Open-Weight Models Challenges US Leadership
GLM-5.3's release underscores China's growing advantage in open-weight AI models despite US restrictions on access to advanced chips for training AI systems.
1
Recent months have seen powerful Chinese AI model releases including Qwen 3.8 Max from Alibaba and Kimi 3 from Moonshot AI, with Z.ai previously stating it used Chinese-made chips from Huawei to train some models.1
The company's earlier GLM-5.2 gained traction among Western developers for coding capabilities approaching leading US models but at significantly lower costs.3
Source: Axios
Vercel CEO Guillermo Rauch reported his engineers tested GLM-5.3 for scanning sites for bugs, stating "Given its lower costs, I expect this to be a boon for defensive security work. It's the new open frontier."
1
Hugging Face used GLM-5.2 to investigate a recent breach after guardrails on US frontier models declined to assist.5
AI expert Nathan Lambert noted the model "looks exceptional, with a somewhat astounding increase in scores" and represents "another step towards the inevitable proliferation of very strong cyber capabilities across the economy."1
The Trump administration is reportedly considering regulatory frameworks for open-source models as their capabilities rival closed counterparts, while the US government now reviews frontier models as part of release processes.1
5
References
Summarized by
Navi
[4]
Related Stories
Chinese AI tools now rival Anthropic Mythos in vulnerability detection as gap narrows
24 Jun 2026•Technology
Chinese AI models close gap with Anthropic and OpenAI as GLM-5.2 tops open-weight rankings
25 Jun 2026•Technology
AI Models Break Out of Testing Labs and Hack Real Systems as Security Pass Rates Stall at 56%
28 Jul 2026•Technology
Recent Highlights
1
AI Leaders Warn UN Security Council of Existential Risks, Call for Global AI Governance
Policy and Regulation
2
AI Hallucination Nearly Sparked US Military Confrontation With Chinese Ship in Middle East
Technology
3
OpenAI AI Agent Breaches Australian Government Healthcare Portal in Historic Hack
Technology
Recent Highlights
Today's Top Stories
Don’t drown in AI news. We cut through the noise - filtering, ranking and summarizing the most important AI news, breakthroughs and research daily. Follow topics that matter to you and stay ahead.
