UK Biobank leak: Health details of 500k people offered for sale on Alibaba
bmj.com
bmj.com
UK Biobank health data keeps ending up on GitHub
https://news.ycombinator.com/item?id=47875843
UK Biobank health data listed for sale in China, government confirms
If they've got anyone with a background in cyber security I can't see it.
https://www.ukbiobank.ac.uk/about-us/people-and-governance/
And then the CEO comes out with:
> We have never seen any evidence of any UK Biobank participant being re-identified by others.
This data contains sex, at least month and year of birth. I can't see any sensible security-oriented technical person coming out with a line like that.
[1] https://gizmodo.com/palantirs-billionaire-ceo-just-cant-stop...
I'd rather see Oracle than a ghoul openly supporting targeting civilians.
Palantir is indeed in many ways just a software vendor but we shouldn’t downplay that they have a much more explicit agenda than most other companies do in seeking government contracts.
2) actually historically, and aspirationally, the US government isn’t supposed to be focused on ideological alignment of its vendors - the current government is anomalous and we shouldn’t normalize this.
yes
In fact the people I have spoken to who have worked on Palantir platform were deeply suspicious of their users treating data with respect, and so built security and immutable auditability as foundational tech.
If anything, the fact the US IC, DOD, NIH, and NHS trust the software with such sensitive and operationally critical data is positive signal
Do you believe these customers don't audit systems/processes that they put their data into?
Yes I believe that it's possible that an organization capable of effective auditing could also leak data
> Are you asserting that Palantir is no different than any other infrastructure company?
For the most part, yes. Palantir is more effective and more "ideological" than most, but in the direction away from your implication that they're vacuuming up data and mixing it across customers
> Do you think there’s any way I would approve of the NHS or NIH using Palantir based on my earlier statements?
No, but the question was whether facts or data adjust your opinion or not, and in which direction if so. Duping one mostly-competent organization on security/privacy posture is much, much harder than duping dozens or hundreds of organizations, including the most security-competent on the planet.
> Is there a reason you’re peppering me with tangential rhetorical questions sort of poking around the premise of what I said
Because the premise of what you said is wrong. The phrase "data vacuum" is clearly meant to imply a fact pattern that just isn't true. The term "unfettered access" is not true either, as data infrastructure companies (Palantir more than most) have significant controls on their exposure to customer data.
The overall implication that people's UK health data would be somehow mixed into a US government effort to kill people is laughably wrong when you actually have to write it out explicitly instead of relying on nudge nudge wink wink.
And yes I do have a Palantir mug actually! Good guess.
Let's get concrete: have you ever actually used Palantir? Ever engaged in contract negotiations with them or set up their access controls on your own data to understand what is or is not allowed, and to what degree you have visibility and control into it?
Sorry buddy but it's you who's speaking in baseless rhetoric here.
They are not an intelligent buyer when it comes to tech.
I don’t get this trend of seeing bad thing happen and then commenting that other bad thing exists and therefore “in fairness” we should downplay it.
Bad things are bad. Comparing them to other things we don’t like doesn’t make them less bad. I don’t like Palantir either but they’re not intentionally leaking health details so this comparison doesn’t even make any sense.
To many, they are. They're leaking information that has been trusted to the NHS to their own databases.
The fact that it's being done under government contract and (arguably) within the law shouldn't immediately make it any less bad.
Of course it should, to say otherwise is absurd
what, the NHS shouldn't have _any_ subcontracting? All data must only be held by sacred NHS monks in a vault somewhere?
As long as palentir are holding the data on UK servers, to modern data security standards, and they have a contract to do so, they should be able to
The point was that they shouldn't use contractors and keep their citizens data private. Whenever they don't do that... that's an issue. Hence the critique.
That was the norm for some time, it's just being eroded over the years and is basically entirely gone at this point
Not just in the UK for that matter...
…says a happy frog who will be as cooked as everyone else.
I personally would like data like this to simply be published, together with a law that says using the data to make personalized decisions affecting those individuals is punishable with life in prison.
Basically, this data is 'opensource', but not for use to decide insurance premiums, job offers, or the contents of news articles.
Like a clean room implementation requirement.
This works well in theory but is basically unenforceable. It's barely possible, if possible at all, to audit how FB or google make ad targeting decisions - but once stuff gets into the fragmented ecosystem of data brokers and market intelligence consultancies all hope is lost.
To say nothing of state actors, like countries who might deny you a visa based on adverse medical info or otherwise use your information against you.
licensing it to researchers allows you to create, monitor, and enforce policies like the one you describe
stealing it does not
As a researcher who regularly deals with such data there is a MASSIVE difference. Yes, I have access to the data but I am restricted on how it can be stored (no cloud), what I can and can't do with it, and for some of it I'm even mandated to destroy it once the research project is over. I have the informed consent of every participant, some of which withdrew halfway throughout the collection without any penalty to them. I also don't need a new law because I'm already bound by existing ones, by the contract I signed when I joined, and by the confidentiality agreement I signed when the project started. While I don't know that the leaker(s) will be identified, the existence of the data itself already calls for legal action while giving a starting point for investigation.
Your suggestion, on the other hand, seems to be "let's put this data out there without people's consent and make companies pinky promise that they won't use it in their black boxes in a way that's virtually impossible to detect or prosecute". Those two things are definitely not equivalent.
When you give O(20000) people you have a 1-0.9999^20000 (high) probability that that will leak anyway (either 1/20000 people not following the rules, or just the accident/attack surface area).
Given the whack-a-mole takedowns, its pretty clear everyone involved knew what was going on.
...until they're inevitably sold.
But once your data has been digitized even if it is under your control the likelihood that it gets leaked is still high. Specially now with AI agents running everywhere, or people just asking AI services for medical advice.
Today the choice for advice is between low quality local AI advice or higher quality advice but lose your data control, the rational choice is probably losing your data control even if if will almost certainly comes back to bite you.
If this is not traceable back to individuals, it would probably good to be made public. But I assume the UK Biobank only gives access to trusted partners since - as we know in our 'data analytics' day and age - with enough general data quantity you can trace back anything to anyone if you have the resources. And the capitalist-surveillance econonmy certainly provides the profit-motive.
If I leak your medical information you confidentially shared it with your doctor that means you are okay with it because you opted in for that?
Or does the scope / details do not matter for others, but only matter for your data.
*I have much better word but I guess I should say it.
Should or shouldn't in general, but THIS one database shouldn't.