Re: [projectvrm] Facial recognition's 'dirty little secret': Millions of online photos scraped without consent

[Guy Jarvis < >]

Hmm,

This is a really interesting aspect of the web, the opt-in assumption/assertion that publication automatically means putting that content into the public domain, unless explicitly stated otherwise.

Robots.txt is a classic example of that approach, ie a spider and index free for all unless explicitly opted-out

Where the clash between cyber and real worlds occurs though is with the reuse of content, so FB were happy to scrape and republish images but object strenuously and litigiously when others copy and reuse/republish whatever FB publish (even though FB isn't even a content creator per se, beyond the functionality of their website).

Guy

On Wed, 13 Mar 2019, 15:05 Benjamin Goering, < "> > wrote:

I think it's a feature that I (or anyone) can download any image on the web.

Sometimes it seems like people want the benefits of publishing things to the world without any of the downsides, but it's a natural tradeoff. Any way of artificially interfering with that is going to be swimming upstream. The only way to win the "private publishing" game is not to publish.

On Wed, Mar 13, 2019 at 7:17 AM Guy Jarvis < " target="_blank" rel="noreferrer"> > wrote:

Rather like FB got started by scraping college year books without consent, there's a pattern here...

On Wed, 13 Mar 2019, 01:05 Dr. Augustine Fou, < " rel="noreferrer noreferrer" target="_blank"> > wrote:

@OliviaSolon: "Earlier this year IBM released a dataset of 1 million photos of people's faces designed to reduce bias in facial recognition software. I was surprised that the pictures were taken from Flickr & so investigated the origins of facial recognition datasets"

https://www.nbcnews.com/tech/internet/facial-recognition-s-dirty-little-secret-millions-online-photos-scraped-n981921

--

Benjamin Goering, Software Producer

bengo.is

@bengo - github.com/gobengo - linkedin.com/in/benjamingoering