-
One of the things I really enjoy about being a FTSAHD
It’s entirely legit to have “make hot chocolate” as a project goal.
-
STUFF TO READ: How to sum up 1700 pages in four words | @webdevlaw on the purposefully intimidating Ofcom “Online Safety” consultation, and where we’ve seen similar before
Sounds familiar: And let’s be clear: it took months for the EC to even consider that they’d made a mistake. Before they did that, they threw in a lot of attacks against the campaigners, the advocates, and anyone who had a problem with a law meant to “rein in the tech giants” but which slapped
-
From 2022: shoshana wodinsky (she/her): “honestly if i had to name the One Big Takeaway from all this shit, it’s that fb employees generally think abt algorithms *way* differently than regulators or reporters do, and tbh i think their way is… the right one ? sorry”
Largely, agreed. It turns out that messy issues are messy. This person has experienced it, secondhand (thread)
-
This new data poisoning tool lets artists fight back against generative AI | “Q: why has everything on the web turned to shit?” “A: proactive vengeance by vitriolic artists…”
The tool, called Nightshade, is intended as a way to fight back against AI companies that use artists’ work to train their models without the creator’s permission. Using it to “poison” this training data could damage future iterations of image-generating AI models, such as DALL-E, Midjourney, and Stable Diffusion, by rendering some of their outputs
-
STUFF TO READ: What the Doomsayers Get Wrong About Deepfakes | The New Yorker
More realism is starting to leak through into the web. Experts have warned that utterly realistic A.I.-generated videos might wreak havoc through deception. What’s happened is troubling in a different way. https://www.newyorker.com/magazine/2023/11/20/a-history-of-fake-things-on-the-internet-walter-j-scheirer-book-review
-
Platforms like Threads are experimenting with Hashtags that are essentially detached metadata, not represented in the {tweet, source of truth}; slight trust & safety problem… #VerbTheOppressedOutgroup
Having experienced “graph search” years ago within [META], I fear that it is solving for “engineers who don’t want to be doing free text search, who think that the Twitter advanced search dialogue is ugly and complex with too many input fields, and who believe that tying user intent to a pre-existing static graph must
-
So it turns out that signing up for an app which explicitly breaks the promise of end-to-end security for an end-to-end-secure messenger, in order to “have your messages all in one place”, is insecure. Who’da thunk?
Of course: Sunbird responded by denying any security issues, and justifying their implementation, doubling down on it being secure. https://texts.blog/2023/11/18/sunbird-security/
-
STUFF TO READ: Twitter’s Former Head of Trust and Safety Finally Breaks Her Silence | WIRED
Interesting interview, not least because reading between the lines of stuff like “getting a semi-effective multi-account detection algorithm in place took years. Years.” — and I know from life experience that part of the reason that’s hard is that REGULATORS HATE COOKIES AND THINK THAT ALL “TRACKING” DEVICE COOKIES ARE FOR ADVERTISING Sigh. I consistently had