EXCLUSIVE: Federal corporations NBN Co, the ABC, and Australia Post have banned DeepSeek from their internal systems despite being exempt from a recent commonwealth ban of the Chinese-based AI chatbot, as the nation's largest telecommunications providers also block the app over privacy and security concerns.

Labor cited national security issues when it banned DeepSeek from federal government devices last week, after Information Age exclusively confirmed a New South Wales government department had banned the app in late January.

Concerns from Western nations have centred around DeepSeek being headquartered in China and using data centres in that country, where companies must comply with data requests from the Chinese government.

While some Chinese companies said they had begun integrating DeepSeek into their systems, Australian organisations which have banned or blocked the AI Large Language Model (LLM) said they had concerns over its domestic use.

NBN Co, the commonwealth company operating Australia’s National Broadband Network, told Information Age it had implemented the federal government’s direction “after careful consideration”, despite not being obliged to do so.

The ABC, which did not respond to a request for comment and was also exempt from the federal government’s direction, blocked access to DeepSeek on its network and devices on 6 February and urged staff to delete the app from work devices.

“While the ABC is not obliged to adhere to this [federal] directive, the ABC has assessed the risks to privacy, security and data protection in the use of this service and are in agreement with the directive,” the broadcaster said in an email to staff.

Australia Post, a corporate commonwealth entity also exempt from the federal directive, said it was “disabling access to DeepSeek AI on all devices, in line with our commitment to data security”.


DeepSeek's mobile apps rose to the top of download charts in late January. Photo: Shutterstock

DeepSeek ‘not worth’ the risk for some

TPG Telecom, which owns Vodafone, said it had banned DeepSeek across its systems and had even recommended workers not use it on their personal devices.

The company’s general manager of tech security, Lee Barney, said the implementation of AI required “a measured and cautious approach from both the public and corporate Australia”.

“As a provider of essential telco services that Australians rely on every day, we are cautious about introducing new and unproven technologies into our systems,” he said.

“Rushing into using new and untested AI models like DeepSeek just to save a few bucks is not worth risking the safety and security of Australians.”

Optus said it had blocked access to DeepSeek for its employees, while Telstra said it had “made the decision to limit access to DeepSeek” for its workers.

Australia’s largest financial institution the Commonwealth Bank said it did not have plans to use DeepSeek, and Information Age understands at least two of the other three Big Four banks are also not using the Chinese LLM.

The University of Sydney and the University of New South Wales both said they were encouraging staff to only use approved AI systems, which did not include DeepSeek, while other universities considered their options.

Australian cybersecurity company CyberCX, which previously issued a warning to local organisations regarding DeepSeek, said it had blocked the Chinese AI from its devices and systems.

"We have done this mainly because of the uncertainty of the data being submitted and who will have access to it,” the company said in a statement.

“There’s also a significant amount of censorship on the platform.”

While US cybersecurity firm Proofpoint said it was not currently blocking DeepSeek, a senior technology director for the company’s Asia-Pacific arm, Steve Moros, said it was important to monitor for leakages of sensitive data from open-source AI models such as DeepSeek, as they had an increased “risk of accidental or malicious data exposure”.

The ABC told staff it had assessed DeepSeek's 'risks to privacy, security and data protection'. Photo: Shutterstock

Chinese firms integrating DeepSeek

While some of Australia’s largest organisations work to mitigate DeepSeek’s potential risks, several companies based in China have added or begun to integrate DeepSeek into their systems.

China’s three largest telcos — China Mobile, China Unicom, and China Telecom — have started working with DeepSeek’s AI model, the country’s Ministry of Industry and Information Technology (MIIT) said on Saturday, as Reuters reported.

Chinese carmaker Great Wall Motor (GWM), which sells vehicles in Australia, told the news agency it had also integrated DeepSeek into its connected vehicle software.

The Chinese government hit back at the Australian government’s DeepSeek ban last week and accused the nation of politicising “economic, trade and technological issues”.

“The Chinese government … has never and will never require enterprises or individuals to illegally collect or store data,” China’s foreign ministry said.

Cybersecurity researchers raise concerns

Several US-based cybersecurity firms said they had identified issues within DeepSeek’s mobile apps, which shot to popularity in late January as the supposed lower cost of DeepSeek's underlying Large Language Model (LLM) caused US tech stocks to dip.

The app was the second most popular free download on the Google Play store at the time of writing (behind shopping app Temu), and the third most popular free download on Apple’s Australian App Store, after dropping behind US competitor OpenAI’s popular ChatGPT.

US mobile security company NowSecure announced on Thursday that it had found “multiple security and privacy issues in the DeepSeek iOS mobile app”, which led it to urge organisations to forbid its usage.

The company said it found the app had transmitted some sensitive data without encryption and stored usernames and passwords insecurely.

Security firm Wiz said in late January it found a publicly accessible database belonging to DeepSeek, which it alleged contained “chat history, secret keys, backend details, and other highly sensitive information”.

The company said it notified DeepSeek of the issue, which was later fixed.

DeepSeek has been contacted for comment, but does not appear to have an official media or PR team.