About | HeinOnline Law Journal Library | HeinOnline Law Journal Library | HeinOnline
LOG IN
LOG IN

GAO-24-107292 1 (2024-03-11)

handle is hein.gao/gaopzo0001 and id is 1 raw text is: Science, Technology Assessment, and Analytics SCIENCE & TECH SPOTLIGHT: COMBATING DEEPFAKES GAO-24-107292, March, 2024 Malicious use of deepfakes could erode trust in elections, spread disinformation, undermine national security, and empower harassers. Current deepfake detection technologies have limited effectiveness in real-world scenarios. Watermarking and other authentication technologies may slow the spread of disinformation but present challenges. ) Identifying deepfakes is not by itself sufficient to prevent abuses. It may not stop the spread of disinformation, even after the media is identified as a deepfake. What is it? Deepfakes are videos, audio, or images that have been manipulated using artificial intelligence (AI), often to create, replace, or alter faces or synthesize speech. They can seem authentic to the human eye and ear. They have been maliciously used, for example, to try to influence elections and to create non-consensual pornography. To combat such abuses, technologies can be used to detect deepfakes or enable authentication of genuine media. Detection technologies aim to identify fake media without needing to compare it to the original, unaltered media. These technologies typically use a form of Al known as machine learning. The models are trained on data from known real and fake media. Methods include looking for (1) facial or vocal inconsistencies, (2) evidence of the deepfake generation process, or (3) color abnormalities. Authentication technologies are designed to be embedded during the creation of a piece of media. These technologies aim to either prove authenticity or prove that a specific original piece of media has been altered. They include: Digital watermarks can be embedded in a piece of media, which can help detect subsequent deepfakes. One form of watermarking adds pixel or audio patterns that are detectable by a computer but are imperceptible to humans. The patterns disappear in any areas that are modified, enabling the owner to prove that the media is an altered version of the original. Another form of watermarking adds features that cause any deepfake made using the media to look or sound unrealistic. Metadata-which describe the characteristics of data in a piece of media-can be embedded in a way that is cryptographically secure. Missing or incomplete metadata may indicate that a piece of media has been altered. Blockchain. Uploading media and metadata to a public blockchain creates a relatively secure version that cannot be altered without the change being obvious to other users. Anyone could then compare a file and its metadata to the blockchain version to prove or disprove authenticity. Sources; GAG analysis (data); let a r~anch esnet'rtoc ad obe -om (images). IGAO-24-107292 GAO-24-107292 Combating Deepfakes Page 1

What Is HeinOnline?

HeinOnline is a subscription-based resource containing thousands of academic and legal journals from inception; complete coverage of government documents such as U.S. Statutes at Large, U.S. Code, Federal Register, Code of Federal Regulations, U.S. Reports, and much more. Documents are image-based, fully searchable PDFs with the authority of print combined with the accessibility of a user-friendly and powerful database. For more information, request a quote or trial for your organization below.



Contact us for annual subscription options:

Contact Us

Already a HeinOnline Subscriber?

Log In

profiles profiles most