How does it work, is it safe to download, UK price

“Experience seamless interaction with DeepSeek’s official AI assistant for free! Enjoy faster speeds and comprehensive features designed to answer your questions and enhance your life efficiently,” the App Store listing for DeepSeek AI promises

APPLE APP STORE | DEEPSEEK AI

What is DeepSeek?

DeepSeek is an AI company based in Hangzhou, China.

Funded by a single hedge fund, known as High-Flyer, the company has sent tremors around the world with the launch of its latest “deep reasoning” model, dubbed DeepThink (R1), that claims to offer the same performance as a rival model from ChatGPT creator OpenAI, despite costing a fraction of the price to develop and requiring much less powerful hardware to run.

DeepSeek currently offers two Large Language Models (LLMs):

• DeepSeek-V3
• DeepThink (R1)

DeepSeek-V3 functions in much the same way as the standard ChatGPT chatbot, offering speedy responses to general knowledge queries, generating never-before-seen text based on your written prompts, rewriting your emails or documents to adjust the tone, and summarising mammoth documents and tables.

Apple recently baked many of these same capabilities into the iPhone 16 and 16 Pro with Apple Intelligence. Likewise, Google Gemini is also capable of manipulating large swathes of text, summarising PDFs into bullet points, and writing everything from poetry to cover letters based on a short written prompt. It’s included in the latest Android phones, like those released by Samsung earlier this month.

This is the sort of AI model that most of us have experience interacting with.

Meanwhile, DeepThink (R1) is comparable to the released-launched ChatGPT-o1 model, which is currently only available if you subscribe to the OpenAI assistant to the tune of either $20 (£16) or $200 (£160).

This model boasts many of the same capabilities, but answers are presented methodically in a step-by-step process — offering insight into how the LLM is thinking about the question, what information it’ll need to draw upon, and the potential pitfalls it’ll need to avoid to provide the correct answer.

This process takes time (you could find yourself waiting up to a minute or more for the final answer) but it’ll help you understand why the AI model has landed on its final answer.

This means you can steer the AI model to a different answer, like correcting a maths question by trawling back through the working to find the error.

DeepSeek AI: See its “deep reasoning” process in action as AI chatbot decides how to answer complicated question

These “Deep Reasoning” processes, which are still in their infancy and rely on human-like reasoning, are seen as a critical step towards achieving Artificial General Intelligence. The latter refers to AI that can match or surpass human cognitive capabilities across a swathe of cognitive tasks.

The team behind DeepThink (R1) has boasted of “performance on par with” the premium ChatGPT-o1 model when tackling complex tasks such as maths, wiring computer code, and natural language reasoning.

DeepSeek UK price

Unlike the comparable ChatGPT-o1 model, DeepSeek-R1 is available to use free of charge. DeepSeek has made all of its AI models open-source model, enabling them to be accessed by anyone.

Unsurprisingly, this has triggered a flood of downloads, with the DeepSeek – AI Assistant app becoming the most-downloaded free app on iPhone in the UK, United States of America, and several other countries.

DeepSeek-V3 and DeepThink (R1) are both accessible online or via dedicated mobile apps for iPhone and Android. There’s no subscription required, just pick the LLM that’ll best serve your needs and then send a written prompt to the AI chatbot, like you would with Google Gemini, ChatGPT, or even Elon Musk’s controversial Grok.

Unfortunately, you might struggle to access DeepSeek today. That’s because the Chinese AI company has temporarily limited new user registrations because of “large-scale malicious attacks” on its services.

Where did DeepSeek come from?

Artificial Intelligence start-up DeepSeek was founded in 2023 in Hangzhou, China and released its first chatbot model later that same year.

Liang Wenfeng, who serves as DeepSeek CEO, previously co-founded a large hedge fund in China that reportedly amassed a stockpile of Nvidia high-performance processor chips used to run AI systems. Since then, sales of high-capacity chipsets to China have been restricted.

DeepSeek claims that its recent AI models have been built using lower-performing Nvidia chips that are not banned in China — roughly 2,000 H800 GPUs from Nvidia, which aren’t the top-of-the-line silicon previously thought to be a necessity when training and perfecting complex AI models.

This revelation has contributed to the recent volatility in the stock market, suggesting that the most expensive hardware may not be essential for cutting-edge AI development. Last month, DeepSeek unveiled an AI model it asserted matched the performance of renowned US firms like OpenAI’s ChatGPT.

Is DeepSeek censored?

Yes, DeepSeek is developed by a team based in mainland China, so it’s subject to the same strict content censorship laws that apply to all companies based there. As such, the AI chatbot is muzzled on certain sensitive topics and historical events.

“We have seen it has been tuned and trained in accordance with Chinese political thought,” Iain Mackay, Director of AI safety at UK start-up Faculty, which works with the UK Government, told The Daily Telegraph.

Asking the latest DeepSeek model for information about the infamous massacre in Tiananmen Square in 1989 didn’t surface information about the violent crackdown by the Chinese military that resulted in at least 200 — possibly thousands — of deaths. Likewise, enquiries about the treatment of Uyghur Muslims in Xinjiang returns a stock response: “Sorry, that’s beyond my current scope. Let’s talk about something else”

Former Apple designer Ben Hylak, who cofounded AI product analytics platform Dawn, shared on X how asking about the so-called Tiananmen Square Massacred caused DeepSeek-R1 to spiral.

Other examples show the DeepSeek AI model start to answer the question before censorship kicks-in and the information is removed and replaced with a stock response.

According to industry-watchers, the censorship isn’t being applied equally, with questions about contentious topics like Tiananmen Square answered in full when the text is typed in Russian. Other queries, like the current challenges faced by the Chinese economy, didn’t have the censorship filter applied and DeepSeek listed factors like slowing growth, a property market crisis, and youth unemployment — not talking points the governing Chinese Communist Party (CCP) likes to discuss.

With the dizzying number of downloads of DeepSeek on the iOS App Store, it raises a crucial question about access to information, freedom of speech, and censorship from foreign governments. However, it’s also worth noting that US companies like OpenAI also muzzle thier chatbots for a number of reasons too.

ChatGPT will refuse to answer questions about a range of topics, including human sexuality, gender, as well as topics that could be seen as erotic or pornographic.

How will DeepSeek AI impact the UK?

Downing Street has declined to comment on the sudden movements of the markets worldwide, but insisted the rise of DeepSeek did not place the UK’s AI sector “at risk of being left behind”.

The Prime Minister’s official spokesman said the UK had “significant strengths in AI” along with “some very strong players in the market” such as DeepMind.

He added: “But our strengths go much further than that and the Secretary of State and the PM set out an ambitious agenda to take advantage of our strengths, but also to ensure that the public sector is harnessing the strengths of AI and indeed the public benefit from AI in their day-to-day lives.”

Prime Minister Sir Keir Starmer pictured during a speech at University College London (UCL) East in east London, where he announced his Government’s AI Opportunity Action Plan that seeks to make Britain a world leader in the sector

PRESS ASSOCIATION

Earlier this month, Sir Keir Starmer unveiled the AI Opportunities Action Plan, outlining the government’s strategy to make Britain an “AI superpower”.

The plan includes creating AI Growth Zones, establishing a National Data Library — a secure database of health and health-relevant data from the NHS and social care datasets, and increasing public compute capacity by a whopping 20x by 2030.

The UK Government says its embrace of AI and other emerging technologices could revolutionise public services, boost economic growth, and improve the lives of everyday people. Key priorities include developing AI talent in the UK, reforming regulation, and driving AI adoption across public and private sectors. The plan has attracted £14 billion in private investment and is expected to create 13,250 jobs.

Why has DeepSeek caused so much stock market volatility?

Being a new rival to ChatGPT is not enough in itself to upend the US stock market, but it’s the purported cost of its development that has caused ripples.

According to reports, DeepSeek is powered by an open source model called R1 which its developers claim was trained for around $6 million US dollars (£4.8 million) – although this claim has been disputed by others in the AI sector – and how exactly the developers did this still remains unclear.

However, anything close to that figure is still substantially less than the billions of dollars being spent by US firms – OpenAI is said to have spent $5 billion US dollars (£4 billion) last year alone.

This apparent cost-effective approach, and the use of widely available technology to produce – it claims – near industry-leading results for a chatbot, is what has turned the established AI order upside down.

As a result, Silicon Valley has been left to ponder if cutting-edge AI can be obtained without necessarily using the latest, and most expensive, tech to build it.

The arrival of DeepSeek has shown the US may not be the dominant market leader in AI many thought it to be, and that cutting edge AI models can be built and trained for less than first thought.

Some commentators have said this may lead to democratisation in the AI research market, as innovation may no longer only be accessible to those with the deepest pockets.

As a result, it could mean more innovation in the sector comes from a broader spectrum of places, rather than just the big names in California. But experts have also said it could have an impact on the world’s approach to China, and in particular the United States, with US President Donald Trump already calling it a “wake-up call” for American AI giants.

Is it safe to use DeepSeek AI chatbot on your iPhone or Android?

DeepSeek AI is currently available as a free app download on Apple’s App Store, where it has recently surpassed ChatGPT and other competitors to become the top-ranked free app, and Google’s Play Store. The latter offers some insight into the performance of the app, revealing that’s already been downloaded over 1 million times.

While this newly-launched chatbot boasts impressive performance comparable to leading AI models like Claude-3.5 and GPT-4o, concerns about data privacy and security have arisen due to its Chinese ownership.

Installing the DeepSeek app will unlock unlimited access to the free AI assistant powered by your choice of either the DeepSeek-V3 model, which boasts over 600 billion parameters, and DeepSeek-R1, which offers the “deep reasoning” model that’s comparable to ChatGPT’s $20 premium subscription tier.

It provides features similar to other AI chatbots, including coding, content creation, and research capabilities. However, it lacks some of the features built into the iOS and Android app for rival ChatGPT, like the ability to upload an image and have the AI assistant answer questions about it — for example, suggest furniture to place in a photo of your empty room, recipes that can be made from raw ingredients in a photo, or fashion advice based on an image of your current outfit. It’s unclear whether these features will be coming to DeepSeek in the future.

However, there are several factors to consider regarding the safety of using DeepSeek:

• Data Collection: The app collects various types of data, including contact information, user content, identifiers, usage data, and diagnostics. While the developer claims this data is not linked to users’ identities, it’s crucial to be aware of the information being gathered.
• Chinese Ownership: As a Chinese company, DeepSeek has raised concerns among security experts about potential data access by the Chinese government, similar to recent controversies surrounding TikTok.
• Privacy Practices: The app’s privacy policy indicates that data handling practices may vary based on factors such as the features used or the user’s age.
• Open-Source Model: DeepSeek’s first release, DeepSeek-R1, is available under an MIT license, allowing for commercial use without restrictions. This transparency may provide some reassurance to users.
• Rapid Rise to Popularity: The app’s sudden popularity and top ranking in app stores suggest widespread adoption, but it’s worth noting that rapid growth can sometimes outpace security measures.

It’s worth remembering that both Apple and Google employ stringent checks before allowing a new application into their digital storefronts. Apple has a fearsome reputation for its safeguards, with every single iOS app and update to those apps verified and approved by an Apple employee working in a division called App Review.

“We’re proud of the store we’ve built and the way we’ve built it,” the Californian company explains on its website. “We created the App Store with two goals in mind: that it be a safe and trusted place for customers to discover and download apps, and a great business opportunity for all developers.”

Likewise, Android apps undergo rigorous security testing before appearing in Google Play, thanks to a system called Play Protect. It scans 200 billion apps every day to ensure phone and tablet owners can download software from the Play Store without worrying about viruses or other issues.

What do the experts say about DeepSeek AI?

The arrival of DeepSeek AI has sparked a wave of enthusiasm and panic across the globe. US President Donald Trump, who announced $500bn in funding for The Stargate Project — “the largest AI infrastructure project by far in history” that he believes will keep “the future of technology” in the United States — just last week, described the arrival of DeepSeek as a “wake-up call” for US giants.

President Donald Trump updates reporters on his thoughts about the emergence of DeepSeek and its impact on the US stock market on board Air Force One

REUTERS

The 47th President of the United States even admitted that DeepSeek could be “a positive”. He told reporters flying with him on Air Force One: “If you could do it cheaper, if you could do it (for) less (and) get to the same end result, I think that’s a good thing for us.”

Senior Portfolio Manager at Pictet Asset Management in Singapore, Jon Withaar said: “We still don’t know the details and nothing has been 100% confirmed in regards to the claims, but if there truly has been a breakthrough in the cost to train models from $100 million+ to this alleged $6 million number this is actually very positive for productivity and AI end users as cost is obviously much lower meaning lower cost of access.

“Is it negative for Nvidia in the short term? Yes, as expectations are sky high on Blackwell (chips)and positioning is long in anything AI supply chain related, but ultimately anything that makes AI cheaper to implement is positive for those selling AI related products and applications and using AI related tools — an ever growing group.”

DeepSeek is accessible from your web browser or via the iOS and Android apps. It boasts two Large Language Models, or LLMs, that can be used an unlimited number of times at no cost

REUTERS

Richard Clode, Tech Portfolio Manager at Janus Henderson Investors in London, said: “DeepSeek appears to bringing some genuine innovation to the architecture of general purpose and reasoning models. Given compute restrictions in China it is not surprising that necessity drives innovation as we saw in Russia in the 1990s when limited access to PCs drove coding creativity and a generation of novel Russian coders.

“As Yann LeCun (Chief AI Scientist at Facebook and WhatsApp parent company, Meta) has noted, this is a victory for the open source model of driving community innovation with DeepSeek leveraging Meta’s Llama and Alibaba’s Qwen open source models. However, DeepSeek is leveraging distillation techniques that are reliant on the work of others.

“How reliant is a key question the market is grappling with currently. Ultimately, we think that creates more questions around the direct monetisation of LLMs (Large Language Models) and open source models than it does about AI capex. This announcement will also draw more scrutiny from the US around chip restrictions as well as the proliferation open source AI models.”

Technology Analyst Ben Barringer, who works at Quilter Cheviot, explained: “DeepSeek’s success will serve to intensify the US/China Artificial Intelligence war, particularly following the recent announcement of the Stargate project in the United States.

“It also provides a wake up call and somewhat of a question mark on how much needs to be spent in order to build a model, and whether quite the level of CapEx we have been seeing is really required. However, bringing the price of model training down is no bad thing as it will help to lower the entry point and this price elasticity could help drive usage and volume.”