Should we ban DeepSeek AI from all Australian devices? Experts weigh in
By Hannah Murphy
Topic:Artificial Intelligence
8m ago8 minutes agoFri 14 Mar 2025 at 7:19pm
A prompt on a screen with the DeepSeek whale logo asking for input.
DeepSeek AI has been blocked on government devices, but experts question if the ban needs to reach the ordinary Australian. (Reuters: Dado Ruvic)
In short:
The federal government banned DeepSeek from government devices last month, citing security concerns.
Some say the ban has not gone far enough, and Australia should follow Italy and South Korea by banning the app on Google and Apple stores.
What's next?
Cyber security experts are divided on the merits of expanding the ban, but say there is a good reason for it.
Cyber security experts say the federal government should consider banning a controversial AI chatbot with Chinese origins on all Australian devices, warning it poses a "unique risk" to national security.
DeepSeek AI was banned from all government devices last month, after a directive from Australian National Security and intelligence agencies found the chatbot put the federal "technology estate" at risk.
The tech was produced in Hangzhou city in China and runs at a fraction of the price of other AI products.
A dark silhouette of a hand holding a mobile phone showing a purple DeepSeek logo and loading screen in front of a Chinese flag
DeepSeek AI's connection to China has raised security questions in Australian intelligence circles. (Reuters: Dado Ruvic)
Australia's national security agencies have warned that because DeepSeek AI is subject to Chinese data collection laws, the app has to give the Chinese Communist Party access to any information it collects.
As a result, security agencies said they were concerned DeepSeek AI could not adequately protect Australian data, could help identify government staff, and could be used against Australia's national interests.
The ban from all government technology attracted criticism from the Chinese ambassador and the Chinese foreign ministry, who called it a "politicisation of economic, trade and technological issues".
But some cyber security experts said they were concerned it does not go far enough.
Luke Roache is the chief commercial officer for cyber security firm CyberCX.
Luke Roache, CyberCX chief commercial officer.
Luke Roache, CyberCX chief commercial officer. (LinkedIn)
"Any piece of connected technology that is produced in, and maintains an internet connection with, an authoritarian state like China presents a unique risk," he said.
Why should the average Australian care?
A chatbot similar to ChatGPT, DeepSeek AI uses conversational language to engage with a user, and can be used to collate information and answer questions with data it gathers from the internet and other users.
According to its privacy policies, DeepSeek collects:
Your email
Phone number
Date of birth
Chat history
Keystroke patterns
Your IP address
Your phone's model
Its data collection policies have attracted concerns from national agencies, as well as those abroad.
Countries like Taiwan, the United States and India have banned the app from government devices, while South Korea and Italy have taken the unprecedented step to also block DeepSeek from its app stores.
Has DeepSeek AI done anything to address the concerns?
The Department of Home Affairs said last month it was hoping to speak with the company to address its concerns around data security.
However a spokesperson told the ABC:
"The department … has not received any formal communication from DeepSeek," they said.
"The government continues to assess Australia's technology security policy settings to ensure they remain fit for purpose."
The department did not respond to questions about whether it planned to follow the lead of Italy and South Korea, banning the app from all Australian devices
"Any further decisions made by the Australian government will be based on national interest and advice from security and intelligence agencies," it said.
However, extending the ban to all Australian devices is already an option being discussed by cybersecurity experts.
The federal government's former national cybersecurity adviser, Alastair MacGibbon, told the Financial Review the ban on government devices was a "good first step" but it needed to go further.
"There is a range of regulators — whether it is the [Australian Prudential Regulation Authority] for the banks or the Department of Home Affairs for telecommunication companies — and they should issue ban orders to those critical infrastructure entities if they can, and if they can't, they should strongly advise that they do the same thing," he said.
Would a public ban do anything?
University of Sydney cybersecurity expert Jonathan Kummerfeld said banning the app on all Australian devices would have little impact.
"Banning the app for consumers wouldn't stop people from using the service by going to the website instead and chatting with it there, so while it would be a signal of concern, the impact may be limited," Dr Kummerfeld said.
"In the government's case, they can also block access to the website, but we don't have a country-wide firewall that could be used to limit access.
A smiling man with square classes, blonde hair and a blue button up shirt
Dr Jonathan Kummerfeld is a senior lecturer in the School of Computer Science at the University of Sydney. (University of Sydney)
"And if we did have a way to limit access to websites there would be concerns about how it could be misused to limit speech."
However, Dana McKay, a professor at RMIT School of Computing Technologies, said she could see the benefit of a wider ban.
"From a national security perspective I can still see a good argument for the ban," she said.
"One of the things that DeepSeek collects is an IP address, which tells them where your device is connecting from.
"There is a good chance that they can use this information to determine when someone is connecting from a government device — and therefore know that they are — at least potentially — doing government work.
Dana Mckay
Dr Dana McKay from RMIT. (RMIT)
"What they search for can then be used to understand what policies are being considered, even under-wraps, for example."
Although Dr McKay said this was not a problem restricted to DeepSeek.
"This is still an issue with other AI companies as well, but we can make laws about what companies do with our data, less so with other governments — and DeepSeek have to follow Chinese law, allowing the government access.
"Another loophole is that DeepSeek collect keystroke patterns, which are like a digital fingerprint — which might allow them to identify a government worker outside of work.
"If that government worker then searches for something really unsavoury, there is the potential for that person to be leaned on."
A stark warning
CyberCX advised government on the ban, and Mr Roache said there was a strong argument for extending its reach.
"Such technologies are subject to direction and control by the intelligence apparatus of that state," he said.
"They can be disrupted or turned off, turned into surveillance tools, or be used as tools of influence to achieve strategic outcomes."
He said the risks of DeepSeek AI were clear and governments should extend the ban to critical infrastructure such as telecommunication networks, hospitals and airports.
"These organisations are already under constant attack by financially motivated cyber criminals and nation-state groups,"
Mr Roache said.
"Given the provenance of DeepSeek and its ongoing connection to an authoritarian state, risks associated with integrity, influence and espionage mean introducing the app into any networks of national significance would be a potential boon for the strategic ambitions of the Chinese state."
Posted8m ago8 minutes agoFri 14 Mar 2025 at 7:19pm
Copy link
X (formerly Twitter)