If you ever thought that AI was some geeky sideline that you could safely ignore, then the scale of the market disruption caused by DeepSeek this week should give you pause.

Yes, it’s an AI tool that operates in the same way as the ones you’ve dabbled with before (ChatGPT or Gemini, for example). But it’s been developed at a fraction of the cost, uses fewer resources and is free to the user. All good, right?

It’s certainly a disruptor and causing waves in the tech community, but, fairly or not, its Chinese origins have caused a certain level of geo-political nervousness.

Frankly, you’re going to need an opinion on this stuff:

Is this really all that important?

More than one investor has called this ‘a Sputnik moment’ – where the realisation that the West generally, and the US particularly, are in a technology battle with a foreign government. It’s not a space race this time, but it is a battle for tech supremacy. Last week, Donald Trump announced a $500 billion joint venture package for the AI industry, and the titans of the US tech industry were all present at his inauguration. All seemed set for the Tech Bros.

Then, on Monday, DeepSeek’s launch coincided with $600 billion being wiped off the value of chip and software company Nvidia. It feels like a seismic threat to the assumed dominance of the Silicon Valley, just as it looked like Musk, Zuckerberg, Bezos et al were the unassailable Kings of the World.

Could it also imperil Keir Starmer’s AI-driven growth strategy? Certainly, the projected £47 billion annual boost to the UK economy looks like small change in the face of such market shifts. But the ground has been shifted by a previously unknown company finding new and efficient ways to do what the tech behemoths already do.

That kind of agile thinking can exist perfectly well in the UK too – big ideas at lower cost can be the playground for the UK tech industry, and in reforming the public sector.

Can I trust the information it gives me?

All generative AI should be treated with a certain human scepticism. They all make mistakes because the Large Language Models (LLMs) they are built on are Large, not Perfect. But the emergence of this from China means that it is subject to local censorship laws which were always thought to be a huge challenge for AI in China. But DeepSeek appears to have been trained on an open-source model, and is able it to perform complex tasks, while also withholding some information, citing that it is ‘designed to provide helpful and harmless responses.’

‘Helpful and harmless’ is a misleadingly pleasant way of explaining censorship, but coming weeks after Meta announced it was dropping its moderation processes on Facebook and Instagram and adopting X’s more laissez-faire approach to content, we should be realistic that we’re not measuring anyone against Western exemplars of Absolute Truth.

Is it Open Source and does that matter?

DeepSeek has released the entire programme for the world to use and adapt, which might be one way for the local censorship issue to be circumvented, if developers outside of China can adapt it to remove these controls.

More widely, such a release can spur further innovation, which is likely to undermine OpenAI (which isn’t Open Source, despite the name) and other US tech giants, which, in part, prompted the stock sell-offs. But it also raises the question as to whether the DeepSeek AI model can be adopted without restrictions and whether uncontrolled and unregulated AI is especially wise for, y’know, the future of humanity.

Is it safe to give it my data?

The release comes shortly after the kerfuffle around the ownership in the US of TikTok, where the issue of how Chinese-owned companies use the data of Western users caused concern. But both TikTok and DeepSeek seem to adopt much the same approach to user data as Facebook, X, Instagram and all the other social media apps you happily use. It’s always been a trade-off between the utility of the app and the importance of your privacy. It’s a pact with the Tech Devil that we all make every day.

With AI apps, the information, data and questions you put into it are also used to train it, so the same common-sense applies to DeepSeek as to all the rest: check the privacy permissions regularly and don’t give it access to your microphone, camera or photos if it ever asks. Don’t upload any sensitive files or images. Just be careful.

Is it better for the environment?

The issues of AI and resources were expertly set out by my esteemed colleague Robin Walker last week. DeepSeek appears to have been launched at a fraction of the cost and using fewer and less advanced chips than it’s US rivals, mainly by using innovative ways of “training” the model. This, in turn, would seem to mean it uses fewer resources and has a lighter environmental footprint.

How true that is and whether that affects future refinements and resilience is yet to be demonstrated. Malicious cyber attacks have already meant restrictions on some new users (but appear to have affected access, rather than been data breaches). But this isn’t a company talking about how it will be investing in huge new data centres (and the associated energy use).

In the end, this is all about the ownership of information – we’re used to Google being our guide to knowledge, but now there’s a whole bunch of other players in the market. How you, as a user or as a company negotiate this space has big implications, and it’s only going to get more complex.

So, should you download the app with the same enthusiasm as everyone else seems to be on the app store? Yes, you should*, even if you just want to get there ahead of your colleagues and competitors. And if you want to discuss the wider implications of this technology for your organisation and its communications, do get in touch.