Dan Romero
@dwr.eth
What is Google's last successful new product — home-grown or acquisition? Define successful any way you see fit.
30 replies
10 recasts
50 reactions
Eliza
@eliza
Probably not the most popular answer but their CSAM detection/removal is getting so aggressive that there was basically a hit piece about how good it is in the NYT yesterday. I could post it if you are interested. As long as their solutions to detection/removal don’t violate privacy rights, I think it’s great.
1 reply
0 recast
0 reaction
deep substrate foliated tomato
@tomato.eth
I'm extremely skeptical of tools like this. https://www.news.com.au/entertainment/tv/current-affairs/aussie-mums-warning-to-parents-after-google-accuses-her-of-posting-child-porn/news-story/ea23eba3ab413c8afd45072fb0d6ed7b
1 reply
0 recast
1 reaction
Eliza
@eliza
Do you have a better suggestion for detection and removal of CSAM/CSEM at scale for google without violating privacy rights or eroding encryption?
1 reply
0 recast
0 reaction
deep substrate foliated tomato
@tomato.eth
I don't have one. I think any detection tool of this type will, by definition, erode privacy rights. If it can scan your private photos for reprehensible content like CSAM, then it can also scan your private photos for anything else, and you have to trust a centralized entity not to violate your privacy.
1 reply
0 recast
1 reaction
Eliza
@eliza
You might be thinking of client side scanning. Even the EU has said no to that for now. What you posted is a story where a 7 yr old exposed his private part on YouTube. The tech detected, removed & reported it. Google has to remove and report by law to NCMEC. In 2022, google reported CSAM 2,174,548 times to NCMEC.
1 reply
0 recast
0 reaction
Eliza
@eliza
Source for that number. The 2023 numbers will be out in January of 2024. A number that high requires some type of tech that detects and removes at scale. Client side scanning is completely different and I’m against that as a solution. https://www.missingkids.org/content/dam/missingkids/pdfs/2022-reports-by-esp.pdf
0 reply
0 recast
0 reaction
