ADVERTISEMENT

News

China's AI chatbot DeepSeek shocked Big Tech, but questions over data security remain

He who controls the semiconductor may not control the universe.

clock

January 29, 2025, 05:49 PM

Telegram

Whatsapp

Back in 2023, OpenAI founder Sam Altman dismissed the idea of building a competitive Artificial Intelligence (AI) model on what then seemed like a meagre US$10 million (S$13.5 million) budget.

"It's totally hopeless to compete with us on training foundation models," Altman declared, fielding a question from the audience at an Economic Times Conversations event.

ADVERTISEMENT

Two years later, Altman's dismissal seems to have been proven completely wrong with the debut of DeepSeek, a Chinese AI chatbot that was ostensibly trained for just US$5.6 million (S$7.6 million).

DeepSeek is thus challenging entrenched assumptions about the AI industry, and shaking up the "Big Tech" world.

It's not just "Big Tech" that is spooked by DeepSeek's debut, however.

Concerns have been raised about the security of the information being fed to the Chinese AI chatbot, and the degree to which it might allow the Chinese state to exploit user data.

The White House, for instance, is evaluating the application on national security grounds, Reuters reported.

Either way, as United States President Donald Trump put it, the application is indeed a "wake-up call" for both the overlords of the tech world, as well as for national security watchdogs around the world.

'You don't need a billion-dollar data centre anymore.': Ho Ching

Former CEO of Temasek Holdings Ho Ching, made a post on Facebook about DeepSeek's disruption, explaining why it is "blowing people's minds".

While AI models like ChatGPT and Gemini require "billion-dollar data centres" replete with expensive graphics processing units (GPUs), DeepSeek achieved comparable results on a relatively lean budget, and with a "few good GPUs".

Nvidia's high-cost H100 chips, for instance, were once seen as vital for generative AI technology, making it one of the most valuable companies in the world:

DeepSeek, however, claims to have not used Nvidia's H100 chips, which are banned in China, to train its AI model.

As a result, DeepSeek's debut caused an upheaval in the stock market and wiped nearly US$600 million (S$810.6 million) off Nvidia's market capitalisation.

As Ho put it, "the 'moats' of big tech companies look more like puddles" now.

Not everyone believes DeepSeek's claims, however. Tesla's chief executive officer (CEO) Elon Musk, for instance, believes that it is "obvious" that DeepSeek is surreptitiously using the H100 chips.

Musk's suspicion is not unfounded, given that there have been allegations of H100 chips being secretly smuggled into China in the past.

Nevertheless, Ho believes that there is no catch to DeepSeek's success.

"Anyone can check their work. The code is public. The technical papers explain everything. It's not magic, just incredibly clever engineering," she wrote. 

ADVERTISEMENT

But, could the chatbot be spying on you?

DeepSeek has become one of the most downloaded apps on Apple's app store in countries such as Singapore and the UK.

Experts who spoke to The Guardian, however, expressed concerns that the Chinese state may be using DeepSeek to gather data or information.

Currently, there is no concrete proof that data submitted to the chatbot is indeed compromised.

All the same, China's national intelligence law states that "all state organs, armed forces, political parties, social groups, enterprises, public institutions, organisations and citizens, shall support, assist and cooperate with national intelligence efforts."

Former Nominated Member of Parliament Calvin Cheng, for instance, posted to Facebook that he deleted the DeepSeek app as it "stores data in China and censors stuff":

The chatbot certainly is quite circumspect when it comes to taboo topics in Chinese politics, such as the Tiananmen massacre of 1989 or territorial claims in the South China Sea, reflecting party lines on these matters.

So, share your data and information with DeepSeek at your own risk.

"An inflection point"

Moving forward, DeepSeek appears set to radically change the AI industry.

Ho explained how DeepSeek's debut sent up 'big tech' companies like Meta, Google and OpenAI:

"DeepSeek did this with a team of less than 200 people. Meanwhile, Meta has teams where the compensation alone exceeds DeepSeek's entire training budget ... and their models aren't as good."

Ho argued that DeepSeek's disruption will make AI development far more accessible, thus increasing competition.

ADVERTISEMENT

Ho added:

"This feels like one of those moments we'll look back on as an inflection point. Like when PCs made mainframes less relevant, or when cloud computing changed everything.

AI is about to become a lot more accessible, and a lot less expensive. The question isn't if this will disrupt the current players, but how fast?"

Top photo from DeepSeek

ADVERTISEMENT

Follow us on Facebook, Instagram, Twitter and Telegram to get the latest updates.

  • image
  • image
  • image
  • image

ADVERTISEMENT

MORE STORIES

ADVERTISEMENT

Events

ADVERTISEMENT