Pay attention to what companies do behind our back and you’ll quickly learn which ones are super creepy.
For example, Mistral Le Chat released the memory feature, but were kind enough to notify me about it. The notification itself also had a nice toggle that allowed me to disable said feature. That was unusually considerate of them.
However, opt-out is still rude when compared to opt-in settings. Think of Debian’s package popularity statistics for example. During installation, Debian gives you the option to enable the statistics or just ignore the whole thing and move on.
Contrast that with Microsoft Copilot that also rolled out the memory feature a while back without telling me anything about it. One day, I just noticed that Copilot is referencing an older conversation, which I find super creepy. That just made me feel betrayed. I already knew that Microsoft is a creepy corporation, just like Google and Meta, so that shouldn’t surprise me one bit. Speaking of Google, better check those gmail settings on again. I’m sure Google can’t stop messing around with them and enabling privacy violating settings from time to time.
2·1 day ago- Its not just true for AI. This shower thought is true for anything with a privacy (surveillance) policy. The sentiment here is also true for just buying video games on steam, or buying a new car.
- You won’t be the only person that has the ability to choose what is done with your data after it is collected. There is an entire distribution chain that you are not privy to after you freely release your data. Some of which might use AI on your data, and some might use opt-out policies that you can’t even decide to opt-out of.
- There are ways to use AI without releasing your private data or agreeing to be surveilled. Self hosting a model locally means that you keep your data on device. The real key here is self hosting and controlling the services you use means that you also control the data distribution.