Ethical Hacking News
Zhipu's new AI model GLM-5.3 is a better bug-finder than Anthropic and OpenAI, according to the Chinese AI company. This achievement highlights the rapid progress being made in AI development and security, and has significant implications for the tech industry.
Zhipu's new AI model GLM-5.3 has been claimed to be a better bug-finder than Anthropic and OpenAI.GLM-5.3 has been benchmarked on the CyberGym benchmark, outperforming other models in cybersecurity challenges.The model's bug-finding capabilities include reasoning across multiple stages of exploitation, forming coherent plans for complete exploitation chains.GLM-5.3 has detected 2,436 vulnerabilities across 269 projects, including 1,097 medium-to-high severity issues.The model's performance on the CyberGym benchmark is a significant achievement, especially in the field of bug-finding.Zhipu's claim suggests China is rapidly closing the gap with the US in AI capabilities.
Chinese AI company Zhipu has made a significant claim in the field of artificial intelligence, stating that its new AI model, GLM-5.3, is a better bug-finder than Anthropic and OpenAI. This assertion has garnered attention in the tech industry, with many experts weighing in on the implications of such a claim.
According to Zhipu, GLM-5.3 has been benchmarked on the CyberGym benchmark, a test of a model's ability to solve real-world cybersecurity challenges. The results indicate that GLM-5.3 outperforms Anthropic's Fable 5 and OpenAI's GPT-5.6 Sol on the CyberGym benchmark. This is a significant achievement, as the CyberGym benchmark is considered a gold standard for evaluating the effectiveness of AI models in cybersecurity.
Zhipu claims that GLM-5.3's bug-finding capabilities are not limited to identifying isolated flaws, but also enables the model to reason across multiple stages of exploitation, forming coherent plans for complete exploitation chains. This level of sophistication is rare in AI models, and Zhipu's claim suggests that GLM-5.3 has made significant strides in this area.
The company has also conducted real-world tests on Chinese codebases, discovering 2,436 vulnerabilities across 269 projects, including 1,097 medium-to-high severity issues. The oldest of these vulnerabilities dates back approximately 40 years, highlighting the model's ability to detect and address long-standing security issues.
Despite GLM-5.3's impressive performance on the CyberGym benchmark, it is worth noting that the model performs worse than western models on other security and coding benchmarks. However, this does not diminish the significance of Zhipu's claim, as the company's achievement in the field of bug-finding is a notable one.
The implications of Zhipu's claim are far-reaching, as it suggests that China is rapidly closing the gap with the US in terms of developing AI capabilities. Anthropic's Mythos, which was previously considered a benchmark for AI models, has lost its advantage to GLM-5.3. This development is likely to have significant consequences for the tech industry, as it may impact the way companies approach AI development and security.
In addition to Zhipu's announcement, other news stories have emerged in the tech industry, including the banning of rideshare operators from offering customers the chance to specify the amount they will tip before a driver accepts a gig in India. Google and Apple have been criticized for their app store monopolies, with South Korea's Communications Commission promising stern sanctions against the companies. Indian tech services giants TCS and HCL have admitted to data breaches, although the company's assurance is that customer data is safe.
Furthermore, Lenovo's enterprise unit has announced a $777 million profit, with AI being a significant factor in the company's success. India's Space Research Organization has issued a tender for the construction of an astronaut training facility, and Microsoft has patched a failed update to its SharePoint software, which is now under a zero-day attack.
In conclusion, Zhipu's claim that its new AI model GLM-5.3 is a better bug-finder than Anthropic and OpenAI is a significant development in the field of artificial intelligence. While the model's performance on other security and coding benchmarks is not as strong, its achievement in the field of bug-finding is a notable one, and its implications for the tech industry are far-reaching.
Zhipu's new AI model GLM-5.3 is a better bug-finder than Anthropic and OpenAI, according to the Chinese AI company. This achievement highlights the rapid progress being made in AI development and security, and has significant implications for the tech industry.
Related Information:
https://www.ethicalhackingnews.com/articles/Zhipu-Claims-Its-New-AI-Model-GLM-53-is-a-Better-Bug-Finder-Than-Anthropic-OpenAI-ehn.shtml
https://www.theregister.com/security/2026/08/17/chinese-ai-company-zhipu-claims-its-new-is-a-better-bug-finder-than-anthropic-openai/5288203
Published: Sun Aug 16 22:30:11 2026 by llama3.2 3B Q4_K_M
