Taiwan has recently joined the ranks of countries banning the use of Chinese startup DeepSeek’s Artificial Intelligence (AI) platform by government agencies due to security concerns. The Ministry of Digital Affairs in Taiwan issued a statement emphasizing the risks to national information security posed by the use of DeepSeek, a Chinese product that involves cross-border transmission and raises information security concerns.
DeepSeek’s Chinese origins have raised red flags in multiple countries, with Italy and several companies already prohibiting access to the AI platform over similar risks. The chatbot has garnered attention for being open source and cost-effective compared to its competitors, but concerns have been raised about its susceptibility to various jailbreak techniques and censorship of sensitive topics.
The popularity of DeepSeek has also made it a target for large-scale malicious attacks, including distributed denial-of-service (DDoS) attacks aimed at its API interface. These attacks, detected by NSFOCUS, were well-planned and originated primarily from the United States, the United Kingdom, and Australia.
Malicious actors have taken advantage of the hype surrounding DeepSeek to publish fake packages on the Python Package Index repository, posing a threat to developers’ systems. The European Union’s Artificial Intelligence Act and the U.K. government’s AI Code of Practice aim to address the risks associated with AI applications and systems by imposing legal requirements and security measures.
The risk of AI systems being weaponized for malicious purposes is a growing concern, with threat actors attempting to exploit vulnerabilities in AI models. Companies like Anthropic are developing new defense mechanisms, such as Constitutional Classifiers, to protect AI models against jailbreak attacks and ensure their security and ethical use.
Source link