Ethical Hacking News
Free Chinese AI models that can hack into computer networks are now almost as capable as the best closed cybersecurity models, raising concerns about the cybersecurity landscape. The emergence of these models suggests that access controls and limitations imposed on closed models may no longer be effective in preventing the proliferation of advanced hacking capabilities.
Free Chinese AI models have been found to be capable of hacking into computer networks, posing a significant threat to cybersecurity. Open-weight AI models, such as DeepSeek, have surpassed closed models in completing complex network instructions. The study's findings suggest that access controls and limitations on closed models may no longer be effective in preventing advanced hacking capabilities. Open-weight models are more cost-effective and provide a strong incentive for companies and developers to use them instead of proprietary models. The emergence of free Chinese AI models highlights the need for companies and developers to be proactive in addressing cybersecurity challenges.
The cybersecurity landscape has undergone a significant shift in recent months, with the emergence of free Chinese AI models that have proven to be quite capable of hacking into computer networks. A new study conducted by Vals, a company that measures AI capabilities, has revealed that these open-weight models are now almost as capable of carrying out complex, multi-step network instructions as the best closed cybersecurity models.
The study, which involved testing various AI models in simulated scenarios, found that the open-weight model DeepSeek was not far behind OpenAI's most advanced cyber model, Daybreak, which is only available to trusted partners. DeepSeek completed 73 percent of the objectives on average, while Daybreak completed 81 percent of the tasks it was given in an average of two tests.
This development has significant implications for the cybersecurity industry, as it suggests that the access controls and limitations imposed on closed models may no longer be effective in preventing the proliferation of advanced hacking capabilities. The US frontier labs, particularly Anthropic, had previously premised the limited release of new models on the idea that gated access would only buy time before advanced hacking capabilities were broadly available.
However, the study's findings suggest that the opposite may be true. The researchers at Vals found that the open-weight models were not only cheaper to use but also provided a strong incentive for companies and developers to use them instead of proprietary models. This is because open models can be run on one's own hardware, making them more cost-effective.
The emergence of free Chinese AI models that can hack into computer networks is a growing concern for cybersecurity experts. While the study's findings are promising, they also highlight the need for companies and developers to be more proactive in addressing the cybersecurity challenges posed by these models.
As noted by Alex Zenla, cofounder and chief technology officer of the cloud security firm Edera, the development of advanced AI models is a cat-and-mouse game between the labs and the attackers. The labs see themselves as moving fast to outpace the open-weight models, to be able to secure vulnerabilities before the open-weight models find them. However, the study's findings suggest that this may be the beginning of a dangerous security landscape.
In conclusion, the emergence of free Chinese AI models that can hack into computer networks is a significant development in the cybersecurity landscape. The study's findings suggest that the access controls and limitations imposed on closed models may no longer be effective in preventing the proliferation of advanced hacking capabilities. As the cybersecurity industry continues to evolve, it is essential that companies and developers be proactive in addressing the cybersecurity challenges posed by these models.
Related Information:
https://www.ethicalhackingnews.com/articles/Open-Weight-AI-Models-Close-the-Gap-on-Closed-Cybersecurity-Models-in-a-Matter-of-Months-ehn.shtml
https://www.wired.com/story/free-chinese-ai-models-are-getting-scary-good-at-hacking/
Published: Fri Nov 6 14:10:06 2026 by llama3.2 3B Q4_K_M
