This post was featured in the Reality Defender Newsletter. To receive news, updates, and more on deepfakes and Generative AI in your inbox, subscribe to the Reality Defender Newsletter today.
Deepfake of the Week
A new deepfake video making the rounds showcases the need to label deepfakes as generative/synthetic media, while also showing a live demonstration of a labeling technology developed by a consortium of technology companies.
What does Reality Defender’s team say?
Unlike other deepfake detection (and labeling) solutions, which predominantly use a single model in an attempt to detect deepfakes, Reality Defender uses an “ensemble of models” approach, measuring deepfakes against known (existing) deepfake models and theoretical (unknown and/or yet to exist) deepfake models. This not only provides comprehensive and more accurate detection than existing solutions; it also creates a way of detecting the latest deepfaking/GenAI methods and potentially those that have yet to exist.
We believe labeling deepfakes is a crucial part of the future, and we’re nearing a day where all platforms, apps, enterprises, and everyone potentially dealing with generated/deepfaked media institutes media labeling and removal protocols — regardless of whether or not legislation calls for it. Due to the advanced and ever-evolving nature of deepfakes, our multifaceted approach allows our platform to outperform any similar detection platforms while consistently remaining several steps ahead of bad actors.
What does Reality Defender’s platform say?
Stanford AI Index Report
Stanford University released their 2023 AI Index Report. The worthwhile read contains information on all facets of AI development, use, proliferation, and monetization. Buried within the text is the above graph, demonstrating the meteoric rise of AI incidents over the years. The latest number in the report comes from 2021, not accounting for last year’s introduction of ChatGPT, rise in popularity of Midjourney, and all related incidents. You can find the full report here.
FurbyGPT
Programmer Jessica Card hooked up a deconstructed Furby to ChatGPT to give the much-maligned ‘90s toy some AI-generated oomph. The results are as terrifying as you would expect.
President Biden Acknowledges “Risks of Artificial Intelligence”
"AI can help deal with some very difficult challenges like disease and climate change, but we also have to address the potential risks to our society, to our economy, to our national security." These were President Biden’s remarks during a recent speech, recognizing the impact weaponized deepfakes and generative content have on a societal level. Ars Technica has the full rundown from the speech.
How A National Public Broadcaster Detects Deepfakes With Reality Defender
Reality Defender's deepfake detection technology helped a national public broadcaster maintain journalistic integrity and audience trust by accurately identifying and preventing the spread of manipulated content.
Get an in-depth analysis of the benefits gained by the broadcaster from Reality Defender’s detection platform in our latest case study.
More News
- Generative AI could raise global GDP by 7%. (Goldman Sachs)
- Novel social engineering attacks are up 135%, partially due to the rise in generative text and content. (ITPro)
- The Guardian formulated a response to ChatGPT-generated Guardian articles. (The Guardian)
- Microsoft’s Edge AI image generator is now open to all. (The Verge)
- Generative AI in fintech alone could create a $6.2 billion+ market by 2032. (Yahoo)
- Canada is investigating ChatGPT’s data collection. (Betakit)
- New generative AI technology makes for some copyright confusion with well-known brands and characters. (WSJ)
- Instant text-to-video media could be the next big step in generative content. (New York Times)
- An iPhone shortcut lets you integrate ChatGPT with Apple’s services. (MacStories)
- Poets think ChatGPT is terrible at poetry. (Vice)
Thank you for reading the Reality Defender Newsletter. If you have any questions about Reality Defender, or if you would like to see anything in future issues, please reach out to us here.
