Other Category
DeepSeek: The AI Revolution We Should Fear?
The rapid rise of DeepSeek and its R1 Large Language Model has sparked discussions about its capabilities, affordability, and the broader implications of using AI models hosted overseas. DeepSeek R1 has been compared to OpenAI’s models, offering similar (and sometimes greater) functionality at a fraction of the cost. However, key concerns around data sovereignty, security risks, and compliance have emerged.
One of the most critical factors when considering AI SaaS platforms like DeepSeek is where and how user data is stored. OpenAI’s data policies allow for storage in “trusted service provider systems in the US and around the world”. DeepSeek, however, explicitly states that user data is stored on “secure servers located in the People’s Republic of China.” For organisations subject to strict compliance requirements, this distinction can significantly impact risk assessments and governance frameworks.
Recent government directives reinforce these concerns. PSPF Direction 001-2025, issued by the Australian Department of Home Affairs, mandates the removal and blocking of all DeepSeek products, applications, and web services from non-corporate Commonwealth entities. The directive cites security risks stemming from DeepSeek’s extensive data collection and potential exposure to extrajudicial directives from a foreign government. South Korea has also implemented restrictions on DeepSeek, and the US is reportedly assessing its national security implications.
For organisations considering their AI strategies, the choice between hosting models like R1 locally, relying on SaaS platforms, or leveraging smaller fine-tuned derivatives requires careful evaluation. Running R1 locally is prohibitively expensive, requiring a significant investment in GPU and CPU hardware, making SaaS-based access the only practical option for most businesses. However, this comes with trade-offs in terms of security, regulatory compliance, and data privacy.
The rise of DeepSeek also brings to light the shifting and tumultuous industry and market for AI. While US firms like OpenAI and Google have dominated the space, DeepSeek has demonstrated that high-performance AI can be developed with significantly lower compute costs. This efficiency has disrupted financial markets, with reports suggesting that Nvidia’s stock took a record single-day hit as investors reacted to DeepSeek’s emergence. The global AI race is accelerating, with new entrants challenging traditional market leaders.
For organisations integrating AI into their workflows, now is the time to implement clear governance policies, robust risk assessments, and user awareness frameworks. Setting clear expectations around acceptable AI platforms will help mitigate risks and avoid unintended data exposure. If your organisation deems DeepSeek too risky but tolerates OpenAI or Azure AI, it’s essential to codify that policy and ensure users are informed.
With AI adoption surging, staying ahead of security, compliance, and operational challenges is critical. If you would like to discuss how these developments impact your organisation’s AI strategy,feel free to reach out.