Australia has banned DeepSeek from all government devices and systems over what it says is the security risk the Chinese artificial intelligence (AI) startup poses.
…
Growing - and familiar - concerns
Western countries have a track record of being suspicious of Chinese tech - notably telecoms firm Huawei and the social media platform, TikTok - both of which have been restricted on national security grounds.
…
An Australian science minister previously said in January that countries needed to be “very careful” about DeepSeek, citing “data and privacy” concerns.
The chatbot was removed from app stores after its privacy policy was questioned in Italy. The Italian goverment previously temporarily blocked ChatGPT over privacy concerns in March 2023.
Regulators in South Korea, Ireland and France have all begun investigations into how DeepSeek handles user data, which it stores in servers in China.
…
Generally, AI tools will analyse the prompts sent to them to improve their product.
This is true of apps such as ChatGPT and Google Gemini as much as it is DeepSeek.
All of them gather and keep information, including email addresses and dates of birth.
…
no shit! that this even needed to be said is concerning.
The APS has pretty clear policies on the use of these tools in general. Some experiments are being run but largely policy is “no, it’s a liability nightmare”
Meanwhile our governments services and tenders practically demand US software and services provided by US companies on US controlled hosting. I haven’t seen any good use for LLMs beyond being an amusement but downloading the Deepseek model to run locally is absolutely safe and local models is all anyone should be using with any data where they have a responsibility, ethical or legal, to maintain privacy and security. And it you are doing things properly and everything is local then Deepseek reportedly has some efficiency advantages that make it worth considering over alternatives.
Preventing exfiltration of Australian data to foreign jurisdictions is absolutely the correct thing to do but block OpenAI and Microsoft and other US companies as well. Once again Australia does whatever its told. I kind of understand when it is the mining barrons or real estate developers given they do at least make some economic contribution to the country. But I have no idea why we suck off US tech bros when all they do is lower our productivity by addicting us to crap products, corrupt our democracy and extort rent from us for the privilege.
Again i ask the question: why am i left with the perception that end users have the ability to acess or install this in the workplace in the first place.
Any IT department worth it’s paycheque would already have everything locked down to hell. I work with a lot of local councils and they’ve grasped this concept
deleted by creator
cause applocker is hard I guess?
Exactly!
All these devices should be locked down, and any apps reviewed and if OK, put on a whitelist.
We got an email at work not to use DeepSeek. And yeah, it’s funny how all the western malware is completely fine
I mean, Copilot and ChatGPT are also banned on my Australian Government device for security, so there’s some consistency at least.
In my department anyway.
western malware isn’t overseen by tyrannical western governments
Yes, it is.
The USA is a tyrannical regime. Their congress is about as meaningful as North Korea’s at this point. They couldn’t even impeach the corrupt criminal.
In fact, on paper, bloody NK already had better seperation of powers than the USA before this election, but obviously it means little because they’re both tyrannical regimes in reality.
As for their malware, NSA TAO have a reputation to uphold. Private corporations aren’t immune, we’ve known about PRISM for over a decade, for a famous example.
With USA’s secret subpoenas we literally can’t know that
And just look at what is happening in the US right now, this statement being arguably true today means nothing about how things will be tomorrow. You can’t put the genie back in the bottle here.
The head of western governments just put off invading Canada for 30 days with 6 hours to spare. All of the big malware companies are loyal to the empire and its military.
You’d have to be mad to put important information into any AI model unless you’re hosting it locally and know it isn’t sending info anywhere (the latter being the hard part to verify). All of the online AI services really should be blocked if departments/companies are taking security seriously.
What about Mistral? https://chat.mistral.ai/chat
I’m not familiar with them, though I did just have a quick browse of wikipedia and their privacy page. From that minimal impression I’d rate their online service as better than DeepSeek (they do claim your data is not used for training, stored in Sweden, encrypted, and deleted after 30 days) but ultimately it’s still got the same problem as other providers in that you have to just believe they’ll actually follow what they say they do. For use with your own personal info this might be an acceptable risk if the company seems reputable otherwise, on the other end of the scale for anything security classified it’d be way too much risk.
Yep, we have https://sh.itjust.works/c/localllama may as well use it, don’t see why not
Especially with LM Studio/ollama able to run on a headless server
You’d have to be mad to
Yes, but at the same time, an astounding amount of people are mad when it comes to tech.
My mate in IT says just this month someone in their corpo office used their work email to sign up to a malicious fake copy of a piracy website. If they were reusing the same password, that could let a hacker into the company account, let alone any other things that employee signed up to on that work email.
That doesn’t even cover the people posting things they shouldn’t on facegram.
That is unfortunately true, for example I find it sadly impressive that one has a decent chance of getting classified info simply by starting an argument on the War Thunder forums…
And yet Copilot is busy burrowing into the flesh of the government like a growing hookworm, a large swathe of big business is simply trusting to Microsoft’s: “Oh no we keep your data entirely seperate and safe. We don’t use it to train the LLM, pinky promise.” Whilst ChatGPT keeps showing up in the hands of the most clueless people, “Oh I gave it all my personal info so it could rewrite my resume. How great is AI!”
I feel like this could be solved immediately and easily, make every privacy breach by any company subject to a fine totalling a single digit percentage of global turnover of the company. So for each privacy breach where Copilot is involved that will be… say… 3 billion dollars. They would yank their “AI Solution” from the local market so quickly you would hear a cracking sound.
good luck getting a Chinese government supported company to pay a fine
How many of these Chinese government supported companies are being provided a veneer of legitimacy by being officially sanctioned to use on state and federal supplied and supported IT resources? Because Microsoft 100% is. Hell they are even getting to supply training materials to government workers on how best to integrate Copilot into their day to day workflows. I am no fan of the Chinese government but I don’t reserve a greater store of trust for US backed Ad-tech companies either and thanks to Five Eyes once one of the aligned governments has your info it’s the same as all of them having it. I have only once interacted with an online LLM, run a few self hosted on my own hardware for probably 3-4 hours and realised that they aren’t worth the power consumption, and really aren’t worth opening a gaping hole into my own privacy. The fact that there are government workers and government organisations who are happily surrendering our data to these companies with no explicit consent sets off more alarm bells than I can express, regardless of the country of origin. And yes I declined the eHealth record and will be doing everything I can to resist digital drivers license because our government is fundamentally untrustworthy and borderline tech illiterate and the IT consultancies they deal with for any IT related infrastructure or services make them look like paragons of virtue and competency.
But that’s just my opinion.
Edit: fixed my spelling, sorry to anyone who gets as annoyed by that as I do 😃
This is the service that is banned, not the model itself, right?
yup, all your personal informal gets sent back to China
