Microsoft Accused of Scraping Users Data
Curious what everyone’s thought is about this bit of news. Hyperbole or a real issue?
If it’s true, it’s extremely serious.
It’s one thing using publicly available information as training data (copyright issues aside). It’s quite another to use personal documents as training data, however much they might be anonymised.
Of course, just because they can in the T&Cs doesn’t mean that they do.
Edit: Microsoft says Word and Excel AI data scraping was not switched to enabled by default (Updated). Microsoft has now refuted the claims. So it seems like it may be a non-story.
Their terms and condition have always suggested that they would scrape customers data.
While it is indefensible for any company to scrape data in a sense Microsoft doing does not concern me as (as I have commented here many many times) I have an aversion to their products and do not have them on my system. Although currently I have Teams and the web version of Outlook but not by choice these are forced on me by organisations that I am currently involved with (and once that is over the offensive programs will be removed).
Seems like a nothingburger. The flag was raised by a Twitter account without any further evidence backing it, Verve caught it.
Extraordinary claims require extraordinary evidence. And this claim is extraordinary because MS would not only infuriate the usual EU privacy watchdogs (that’s a GDPR violation without explicit user consent) but also its corporate customers which are their biggest income. I don’t say that MS are saints, nor that they haven’t tried schemes like this before, but just saying that without any other proof is simply click-baiting.
And people wonder why I use Office 2021 & no 2024 LTSC with a script to disable all telemetry and a software firewall to block all connections.
Their response on X doesn’t sound like a refutation to me…
“We do not use customer data to train LLMs” isn’t a refutation?
2 Likes
They wouldn’t be able to do this with Corporate customers. MS bends over backwards to not piss them off. It would be consumers who would get hit by this, especially those who take free services.
2 Likes
or SoftMaker Office with no telemetry…
1 Like
It only covers LLMs but anything else could be game.
Exactly this. we have them at our corp and our info is secured behind a business service agreement (BSA)
The home customers are the ones getting data mined here, mark my words.
1 Like