AI Shield
End to End encryption will become a moot point as far as privacy is concerned once AI becomes integrated into OS. Microsoft has made it clear that it wants copilot to be "An AI companion for everyone" and "see what we see and hear what we hear", NVidia wants to copilot everything. It's currently opt in, but if history has taught us anything, that won't last as soon as MS has finished bringing the frogs to a boil. So, even if it's end to end encrypted, it will be as private as someone standing over your shoulder watching everything you do, except it's someone with perfect memory, that's likely sending that info to their boss (hint, that will never be you). Seeing as MS has already been caught unlocking/scanning encrypted files by trying passwords you had saved with them in One drive, sharepoint and teams, this suspicion is neither unlikely nor alarmist. In fact, there's a good chance it won't require you interacting with Proton apps, it'll just help itself to the data to "improve your user experience" or something. So to maintain privacy Proton would need to develop something that stops AI from being able to see or read the info, which AI will be able to do in any manner that you (the user) can meaningfully interact with or view it. A product that sandboxes and blocks AI from viewing your Proton data when accessed would be nice. I'd assume Apple and (definitely) Google native apps will benefit from this layer of privacy and security also. Thank you to the Proton team, your commitment to privacy and security and all the hard work that you do
Tony the Geek
shared this idea
-
Edward Evans commented
Scrolling through the comparison table was actually what gave me the clearest picture of how fragmented this AI niche has become. Different platforms can offer similar basic functions while approaching pricing, processing modes and additional features in completely different ways. I used the CPA.LIVE article to make sense of those differences. The https://cpa.live/en/ai-tools/nude-ai-tools/ guide lists various systems and describes their photo engines, video presets, quality levels and access models. I noticed that some services are focused mainly on photographs, while others put considerable attention on video generation. There are also differences in how free access works. A platform may offer a demonstration or several free credits, while continued use requires tokens or a subscription. That distinction was useful to understand before evaluating any service. I also read the safety section because the technical side is only part of the picture. Since many of these tools process files on external infrastructure, users should consider whether images are temporarily stored and whether technical or billing logs are retained. The article recommends checking privacy policies and avoiding sensitive photographs. It also emphasizes that images involving other people should not be processed without appropriate consent. After reading the comparison, I found it much easier to understand what questions should be asked about an AI image service before deciding whether it is worth exploring further.