DeepSeek: Diving Deep into the Artificial SpAIce Race

DeepSeek ravaged headlines and the tech-heavy Nasdaq 100 last week, when its R-1 models reasoning capabilities were made available to the public.

Up until last week, no Generative AI platforms were apt to compete with the strength of the US led industry, that is until China made its own cheaper AI model – supposedly using less than US $6 million to train it (in comparison to OpenAI’s $100million). What’s more – America’s stringent regulation on the export of AI superconductor chips seems to have served no hinderance to the Chinese powerhouses’ ability to produce a comparable AI machine. Despite all this, DeepSeek performs equal to, and in some benchmarks (namely, math, Chinese, and coding) better than its proprietary predecessors – which the American Tech Industry has attributed largely to efficiency improvements.

Founded in 2023 by Liang Wenfeng, the almost unbelievable learning speed exhibited by the AI model has some folks comparing it to “the student who claims they got the top test score in the class after cramming for 15 minutes”. Others are stating that a comparable reference to the cost of developing such a platform is like selling a brand-new iPhone for as little as $50. Needless to say, DeepSeek has more than just the Tech industry turning heads, and with a host of conspiracies surrounding the celebrity-like platform, our Technology lawyers think this is a good opportunity to clear the air around the use of AI generally.

How cheaper AI impacts the market

While the initial market reaction appeared to be shaken to its core-processors, the biggest Tech CEOs in America turned the worrying narrative on its head. CEO Mark Zuckerberg, stated that despite what DeepSeek had accomplished with relatively limited resources, Meta intends to spend a further $60-$65 billion US on capital expenditures this year, with Microsoft announcing $80 billion for AI data centres. The Tech industry now seems grateful for DeepSeek’s introduction and believes that DeepSeek’s resourcefulness will only lead to further growth of AI infrastructure.

Australian AI policy and DeepSeek

Owing to the expected growth of the AI industry, and naturally the increased adoption of AI tools in businesses, such businesses will need to be mindful of the risks posed by its use – especially with further regulatory introductions anticipated for Australia.

DeepSeek privacy concerns

One of the overarching concerns currently relates to DeepSeek’s terms of use, which essentially allows the AI platform unfettered access to the users information, including keystroke patterns and the ability to harvest data from any other application on that device. By downloading the App, a user is permitting the collection and use of all the information on that device (including if you access your email accounts from that device, potentially confidential information).

By agreeing to its T&Cs, DeepSeek will gain access to the data and information, which may see those working in professional services in breach of their professional standards and obligations to their clients. This may also result in a breach of confidentiality obligations in contracts with other businesses and possibly even loss of professional privilege!

Downloading DeepSeek to a ‘work device’ may constitute a data breach for your business. Its own terms of use and privacy policy state that any user of DeepSeek consents to this collection of information and disclosure of it for any purpose and to any third party. This has resulted in many entities banning DeepSeek on work devices, including the Australian government and the US Navy.

DeepSeek and AI’s impact on business

DeepSeeks entrance to the AI race has solidified the notion that AI is here to revolutionise business. Those who don’t adopt AI will almost certainly fall behind. As AI continues to evolve, businesses must navigate the ever-changing and intricate web of AI policies, privacy regulations, and ethical considerations. The handling of data must align with a business’s internal processes and procedures, and employees also need to be aware of how their actions and use of generative AI can impact on their employer (and themselves).

What can businesses do in the wake of rapid AI development?

If you are operating a business which uses or intends to use Generative AI in relation to your operations, or you want to allow your employees to utilise it on their ‘work devices’, you must be aware of the potential implications for doing so. Additionally, you should put in place policies and procedures for dealing with generative AI, including a procedure for adopting new AI and/or discarding old AI systems.

If you do have concerns about your use or intended use of Generative AI, Macpherson Kelley’s Technology team has extensive experience with the development of personalised AI Policies and is ready to assist your business in safeguarding the privacy of its operations. If you are otherwise concerned about your organisation’s exposure to compliance with current law or impending regulations, or want to discuss the above matters further, please get in touch.

The information contained in this article is general in nature and cannot be relied on as legal advice nor does it create an engagement. Please contact one of our lawyers listed above for advice about your specific situation.