The appearance on the market of DeepSeek, the Chinese Large Language Model (LLM) available in Open Source, has prompted two US Congressmen to propose legislation to ban it from Government devices to protect national security.
Before jumping to any facile conclusions, a few dogmas need to be dispelled. The first concerns the cost and time required to generate an LLM. If it is true that the development of DeepSeek did not enjoy Beijing’s direct support in respect of privileged access to the hardware and energy needed, then it is no longer true that billion-dollar investments are necessary to compete in the market. It therefore follows that it is feasible to increase the number of potential players in the industry and to create a genuine international competition regime.
The second dogma challenged by DeepSeek is that the AI industry is firmly in the hands of the US, which controls both software and hardware technologies — from graphics cards to the technology needed to build ever thinner chips and ever more powerful processors. The technology ban imposed on China by Washington has led the country to develop alternative approaches in both theoretical research and dedicated AI hardware, such as that produced by the controversial Huawei system.
The public availability of DeepSeek in the form of a downloadable app on smartphones and platform had an impact on the financial market that hurt the market value of Nvidia, the near-monopolist manufacturer of GPUs and AI software development environments. The fluctuation was, however, brief, and its shares recovered almost immediately, but it was a clear signal of what could happen in an industry in which value volatility is heavily influenced by the dissemination of information or, rather, how investors perceive the information disseminated.
If the news about DeepSeek’s higher cost-effectiveness affected the stock market, the Chinese startup’s choice to release the model in Open Source (that is, allowing its use by anyone without claiming royalty or rights payments) attacks the real market. Perplexity, one of the most active American AI companies, recently announced that it is using DeepSeek to deliver certain services. It is true that everything ‘runs’ on American systems, no data are sent to China, and no one except Perplexity has access to the model. But this does not alter the fact that a single company has been able to enhance its services without having to pay licensing fees to competitors developing similar models. If other companies follow Perplexity’s lead, the industry’s Big Techs will inevitably face domestic competition capable of taking market share and disrupting the public release schedule of new technologies.
Is DeepSeek the new TikTok? In itself, DeepSeek’s diffusion in the West could be framed as part of a ‘normal’ competitive dialectic between industry companies. But given the strategic nature of AI, it is clear that this reading would be very reductive. If, in fact, after the first technical trials the Chinese LLM diffusion mechanism were to come into full swing, we would be faced, willingly or otherwise, with something that looks very much like an act of economic and technological warfare.
Nevertheless, the announced submission of a bill to ban the use of DeepSeek from Government devices is not based on these concerns, but rather on the fear that the app installed on smartphones and tablets could provide user information to the Chinese Government.
Although the two events are not entirely overlapping, it is quite clear that the call to ban the use of the app is based on the same assumptions that led to forcing the forced sale of TikTok. The significant difference between the two cases, however, lies in the fact that in the case of DeepSeek the need to protect national security is stated in an essentially precautionary manner. Where, in other words, against TikTok the US intervened when the application had already been in place for years, with DeepSeek the US legislators seem to wish to apply a precautionary principle by imposing a ban on use even before a problem could manifest itself.
In this regard, it should also be mentioned that, as the judgments of the courts that have dealt with the case have clearly stated, the reasons for the forced sale of TikTok were based on Government statements that the judges did not see fit to challenge, while in the case of DeepSeek no similar positions appear to have been taken.
With respect to national security, it is not yet possible to know whether and to what extent DeepSeek’s ban will be ordered; but the mere fact of ventilating the hypothesis fits the pattern of similar choices already made by US administrations against Huawei, DJI and TikTok. All of these decisions are united by the tendency to view control over a technology by a foreign state as a possible threat to domestic survival regardless of the material employment of a product or service that that technology uses.
There remains an irreconcilable contrast between security protection and global techno-economic development. While the principle of ‘better safe than sorry’ applies in the context of national security, deploying preventive measures based on risk assumptions instead of verifiable evidence raises questions not only about international law and competition, but also regarding the future of global technological governance.
If the precautionary principle becomes the tool for systematically excluding innovations from certain countries, there is a danger that an increasingly fragmented technology sector, with the creation of separate and incompatible digital ecosystems, could be on the horizon. Such a scenario would not only hinder scientific progress and international cooperation, but could also prove counterproductive for US companies themselves, which would lose access to innovative models and solutions developed outside their own borders.
Andrea Monti and Raymond Wacks are co-authors of Protecting Personal Information: The Right to Privacy Reconsidered; COVID-19 and Public Policy in the Digital Age; and National Security in the New World Order: Government and the Technology of Information.
To join in with the discussion please make a donation to The Daily Sceptic.
Profanity and abuse will be removed and may lead to a permanent ban.
Even if you ask a deep question the answers tend to be rather cagey. The limits are the limits of analysis compared to inspiration. If you are impressed by AI then you probably have very limited knowledge of literature and culture. The whole spirit of literature is the rejection of AI. There is always a sense that the writing down involves a sacrifice and loss.
Unfortunately, being an extremely stupid dog with paws unsuited to touch screens I haven’t yet worked out what the real utility is of (misnamed) AI or indeed tiktok unless it is reduce the youth of western countries to utter imbecility, an objective it seems to be achieving satisfactorily well.
LLMs seem to be reasonably ok at digesting large volumes of information and summarising it accurately and coherently in response to questions. They are also ok at generating images though that’s not really my field.
It depends how you define “intelligence” but AI is perhaps an unfortunate term.
I am similarly uninterested in the (misnamed) AI.
I honestly don’t think we’re missing much, Jack the dog.
I speak as a professional in the IT field.
But please do not take my word for it. It could be amazing. In the meantime, I need to put another log on the fire.
Your enemies lie within. It is just stupid to pretend otherwise and that the Chinese are looking to attack you. For God’s sake you only have to have lived in a western country in the last thirty years. You destroyed yourself.
Never use touch screen technology. It is good to have a touch screen because it tends to be durable but please never actually touch anything on the screen.
Why?
It is stupd for many reasons. They wil easily beat you in maths and physics exams and will generally have more vitality. But they aren’t trying to destroy the West. Their very existence is predicated on trying to be a good emulation of the west in certain ways. If you go all over Asia they love American culture and the American people. It is assumed that the Iranians hate America. This is the opposite of the truth they are the most pro-American culture in the world. They assume that Iranians hate Jews when Iran gave refuge to the Jews for 2500 years and still the real capital of Jewish power is Tehran and Jewish culture is in large part Persian culture, You shouldn’t listen to the crap they pedal.
If deep seek is truly open source the sureky it should be possible to identify any malign intentions from analysis of the source code.
From a brief search there seems to be some dispute as to whether it’s truly open source. I don’t know enough about how LLMs work to have a view myself.
Ta!
They don’t want you. They don’t care. They have better things to do.
I am a lot less worried about DeepSeek (i.e. not at all, I don’t use it, I don’t know what it’s for) than I am worried about what Google, the NSA, CIA, MI6, GCHQ et al already think they know about me.
Who paid the developers is the unanswered question? It has to be the CCP!