Here's an #AI tool that arrives just in time: The PRISA media group has created an audio #verification and #deepfake detection service for voices in Spanish.
Their goal: Support #journalists (e.g. when covering election campaigns) and foster trust in the face of #disinformation.
Remarkable: The creators also address the problem of powerful people claiming their voices have been cloned after saying something embarrassing.
(1/2) Can AI filters help reduce the harmful emotional impact of gruesome images that journalists and human rights investigators need to check as a part of their daily job?
A new paper, researched and written in coop with the Centre for Research & Technology Hellas (CERTH) in the scope of the MediaVerse project, sheds some light on the issue.
(2/2) A key finding of the user study: AI-based "drawing style" filters actually work pretty well. They seem to reduce negative feelings of viewers by more than 30% while preserving almost the full interpretability of an image.
Link to the paper by Ioannis Sarridis, Jochen Spangenberg, Olga Papadopoulou, and Symeon Papadopoulos:
(1/2) By now you've probably heard about #Sora and the latest wave of #AI-driven text-to-#video tools. It's all quite spectacular, but far from perfect. If you take a closer look, it's still possible to detect synthetic content. Essential pro tips: Play everything in slow motion. Consider real world physics – and logic.
To identify and understand synthetic content on next generation social platforms and to provide professionals with tools and guidelines for the new media landscape they're facing – that's the in-a-nutshell mission statement of AI CODE, a new media R&D project co-launched by DW in December 2023.
"Higgins (...) thinks of himself and his @Bellingcat colleagues as 'an open community of amateurs on a collaborative hunt for evidence... an online collective, investigating war crimes and picking apart disinformation, basing our findings on clues that are openly available on the internet.' As such, he can get going on his day’s work while eating breakfast and getting the kids to eat theirs."
"DF2023 is a dataset for image forgery detection and localization. The training and validation datasets contain 1,000,000/5,000 manipulated images (and the ground truth masks)."
In English and German, there will be updates on the cat and mouse game of faking and debunking news content + discussions with other journalists, media creators, and OSINT people + demos of useful tools. You can also have a 💬 + a ☕ with our colleague Anna Schild.
The assault by Palestinian militant group #Hamas on #Israel and Israel's counter attacks have led to a great extent of human suffering – and to a flood of #disinformation on social media.
DW's #factchecking team has produced a video that documents several viral fakes and explains how to debunk them.
Mastodon allows you to verify yourself as the owner of your website. If your website is well-known, then this will in effect verify your identity as well.
Here's how to verify you are the owner of a website:
I've had my share of being impersonated online, so if there is an "official" authentication system available, I take the opportunity (but not Twitter and Meta, they rejected me multiple times).
So, I guess before this week ended, there's something good that happened. It at least lifted my soul.
We have a matrix space
We have a chat room for the smallest of talks!...