\
insight
\
\
Insight
\
Reality Defender
Team
AI and disinformation experts spent the last few years warning the world that a viral deepfake shared on social media would one day cause a material impact on financial markets.
That day was Monday, when the above photo of an explosion near the Pentagon was shared on twitter by verified (read: paid blue check mark) accounts as “real” and seen by millions of users. This caused a small flash crash, with a temporary loss of billions across financial markets due to the perceived fear of an attack on American soil.
Our platform found the image to have an over 99% probability of being a deepfake. Doing this with the naked eye is nearly impossible, but if you look close enough, the railing has inconsistencies and even floats on the sidewalk, while the smoke is overly smooth compared to the building behind it. AI analysis found artifacts throughout the image that were in line with diffusion-based generated images.
Reality Defender Co-Founder and CEO Ben Colman went on CNBC’s Last Call to talk with Brian Sullivan about this unique case, Twitter’s failure in preventing its widespread release, and why a multi-billion loss of stock market value is the least worst thing to come.
Sam Altman, CEO of OpenAI’s for-profit arm, spoke last Tuesday before a Senate judiciary committee on the regulation of AI (which Altman claims is “essential.”) “We think that regulatory intervention by governments will be critical to mitigate the risks of increasingly powerful models,” said Altman during his testimony. At the same time, Atlman proclaimed the release of OpenAI and other tools “vastly outweigh [their] risks,” though the company does test and employ safety measures before releasing new systems (like GPT4) to the public.
Altman (along with other members of the OpenAI team) also wrote on the coming worries about a “superintelligence” and the need for establishing a governing body to regulate it. The whole post is a fascinating read, including how the company feels that such a superintelligence could surpass AGI, the “holy grail” of artificial intelligence.
Committing fraud with deepfakes in Taiwan now carries a prison sentence and/or a fine after new legislation passed on the matter this week. Read the full report on the amendment to the country’s Personal Data Protection act in the Taipei Times.
As the U.S. figures out potential laws and regulating bodies for AI, U.K. Prime Minister Rishi Sunak is looking to monitor and reign in control of AI "with guard rails in place." The Guardian has the full story on the PM’s comments, which are more in line with the dangers of applications like ChatGPT and the possible redundancy-related ramifications.
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.