My incredible colleague, Antigone Davis, Head of Global Safety at Facebook, has posted about our test in Australia to combat Non-Consensual Intimate Imagery (NCII).https://newsroom.fb.com/news/h/non-consensual-intimate-image-pilot-the-facts/ …
-
Show this thread
-
Some salient points: 1) We already have a mechanism for victims of NCII to report images that are posted on our products. This test is intended to help those victims who are being blackmailed by an abusive partner or criminal and who want to take action.
2 replies 8 retweets 27 likesShow this thread -
2) This test was put together with the help of the Australian eSafety Commissioner and several advocates and NGOs with deep experience with NCII and partner abuse. Check out their statements in the post.
3 replies 7 retweets 22 likesShow this thread -
3) We are aware that having people self-report their images carries risk, but it's a risk we are trying to balance against the serious, real-world harm that occurs every day when people (mostly women) can't stop NCII from being posted.
10 replies 15 retweets 51 likesShow this thread -
In recognition of that risk, we have taken the steps we can to protect this data and to only retain non-reversible hashes. To prevent adversarial reporting, at this time we need to have humans review the images in a controlled, secure environment.
5 replies 9 retweets 31 likesShow this thread -
4) We are not asking random people to submit their nude photos. This is a test to provide *some* option to victims to take back control. The test will help us figure out how to best protect people on our products and elsewhere.
4 replies 8 retweets 33 likesShow this thread -
A personal comment: this situation is a pretty good example of a problem I talked about at Black Hat this summer. Our greater infosec/privacy community, including the media, has trouble talking about imperfect solutions to serious problems.
8 replies 62 retweets 172 likesShow this thread -
When there was a burst of ill-informed press around a crypto issue this year, there was an ability to self-correct, since the external experts who set the tone of the conversation understand crypto.
3 replies 4 retweets 27 likesShow this thread -
The number of people outside of the big platforms who have to deal with abuse like this at scale is actually minimal. I think this is a good demo of why we need to continue to talk about these problems publicly to build understanding and trust.
3 replies 12 retweets 68 likesShow this thread -
A quick note to everybody who says "calculate the hash locally": A) Photo fingerprinting algorithms are usually not included in clients to prevent the development of circumvention techniques. B) Humans need to review to prevent adversarial reporting.
16 replies 24 retweets 59 likesShow this thread
I'm not sure I buy B) as a significant problem. You could very well structure it so that the human review happens the first time such a hash is found in the wild.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.