So i eventually solved my own problem after some searching and inferring. It seems kind of obvious now but if you have your app's Assign to set to "All Desktops"it causes this behavior. This fix is to set assign to, to none and everything seems to behave as i would expect.
And it showed that KN95s are NOT perfect at blocking COVID. This is why the general public now distrust scientists, because the media presents it to them inaccurately and tells them results the study never found. Science is about precision. The study did not find all N95s where nearly perfect at blocking covid, that's flat out inaccurate.
Also let's be really precise here. I'm not conflating anything and i think you don't realize you are.
1) KN95 and N95 use similar filter materials they both have to filter 95% of particles down to 3 microns, the difference is to do with certification N95 is a US certification and KN95 is a Chinese certification. So the material is essentially the same between them. The fits tend to be different.
2) What i'm getting at is the title said all N95s, that's not what the study found, it found that a specific fit style "duckbill" was "Nearly Perfect at Blocking COVID" the non duckbills did not.
So to be accurate what the study did NOT find which the title suggests is that all US certified masks that filter 95% of particles down to 3 microns were nearly perfect at blocking COVID.
Read further down in the study. It discusses exactly what i'm referring to. The title says "N95 Masks Nearly Perfect at Blocking COVID" that's not what the study found. It found that duckbill N95s were nearly perfect at blocking COVID.
Read further down in the study. It discusses exactly what i'm referring to in that air was pushed out the sides with KN95 masks and one of the reasons cloth masks performed better then KN95s was a better fit.
The title is misleading. The study found duckbill N95 masks "blocked 99% of large virus particles and 98% of small ones". This is important because the issue with masks was always to do with fit and longterm ware habits, not that N95 material was ineffective at blocking viral particles.
I built a prototype along these lines as well. I was more concerned with actually giving sensitive data to ChatGPT even through their API. So it used a local LLM to scan the text first and replace sensitive info before sending on to ChatGPT. Unfortunately when i tried to interview customers to see if it was a viable product idea i got very little interest. I ended up building it into a chat app i was going to release but that’s also in limbo since ChatGPT released their desktop app. I could see these 2 things being combined, i just don’t know how big the market is for a product like that.
https://www.texasattorneygeneral.gov/news/releases/nationwid...