We are a trusted resource for raising public awareness on the issue of algorithm-based fabrication. Through a network of partnerships and the tipline, we aim to combat generative A.I. based misinformation. We are structured to leverage the expertise of academics, researchers, startups, tech platforms, fact-checkers, and journalists to help the public discern the real from the fabricated.
The DAU won the award for the "Most Innovative Collaboration" at GlobalFact 11
SYNTHETIC MEDIA
What are Deepfakes?
What we check for ?
A.I.-Generated Content
A video/audio created using artificial intelligence that depicts an event, a person, or an interaction between people that never actually occurred or that alters the reality, which may or may not mislead people.
Manipulated or Altered Media
Manipulations made in video/audio using simple editing software or sophisticated technology (A.I.) to create a false or potentially misleading narrative. Examples of this type of manipulation could include altered/edited videos that are created to add/remove important context, images, and videos that are wrongly labeled.
Deepfakes
A deepfake refers to any synthetic media that includes fictional creations about a living or dead person, or a non-existent person, for example, a political party worker who doesn’t exist. Some common examples of deepfakes include swapping the face of a person (dead or alive), or cloning their audio to an extent where the generated content shows them doing something they didn’t do or saying something they didn’t say.
Cheapfakes/Shallowfakes
A cheapfake refers to a video/audio that has been created or manipulated using relatively less sophisticated technology like Photoshop or simple editing software, however, in some cases the production of that piece of media could involve limited use of A.I. This includes slowing down, speeding up, reassembling, re-contextualising, and editing footage to create a false or a potentially false narrative; such techniques are also used to create content that is satire or parody.
* These are living definitions meant to guide the DAU's operations. We are open to feedback and suggestions and will keep revisiting these definitions to update them based on what we see and the rapid advances in A.I. technologies.
How we work
OUR PARTNERS
Fact Checkers
Detection and Forensic Partners
Technical and Research Partner
Media Organisations
Supported by
OUR PHILOSOPHY
Philosophy of the DAU
Transparent
Transparency is a must for credibility, trust, and accountability. From processes and methodologies, to working with the public, and building open and honest collaborations with partners, our operations will be guided by transparency.
Collaborative
The Deepfakes Analysis Unit will serve as a bridge between fact-checkers and experts working on generative A.I. detection, in turn fostering better knowledge exchange and resource sharing than conceived before to address misinformation- at- scale.
Open Source
Openness is a commitment. Anyone can use, change and share our tools. We welcome everyone to inspect/enhance the source code of Deepfakes Analysis Unit, created by Tattle, for their unique purpose.