New York politician says 'most dangerous' uses of AI can be solved with 1990s technology that made online banking possible

Applications of AI


New York politician says 'most dangerous' uses of AI can be solved with 1990s technology that made online banking possible

New York politician Alex Boas said the problem of deepfakes (highly realistic photos generated by artificial intelligence) is solvable, and the solution is not to train humans to spot deepfakes, but to incorporate the technology that saved the early internet and laid the foundation for online payments. Boas, a Democrat who is currently running for Congress in Manhattan's 12th District, said a return to cryptographic tools could alleviate the “crisis” of AI-generated misinformation.“Can we figure out deepfakes? Because this is a solvable problem and one that I think most people are missing the boat on,” Boas, who previously worked as a data scientist and federal private business leader at Palantir, said on a recent episode of Bloomberg's Odd Lots Podcast (via Fortune). He proposed that the technology that made consumers trust online banking in the 1990s could be applied to verify the authenticity of every image, video, and audio clip online.

how digital certificate Internet banking “problem” solved

According to a report in Fortune, in the 1990s, some believed the Internet would never be secure enough for financial transactions, but the proliferation of HTTPS and digital certificates has changed that, allowing users to verify that a website is authentic and secure.Boas argues that a similar “trust but verify” architecture is key to neutralizing deepfakes. “It was a solvable problem. Essentially the same techniques can be applied to images, video and audio,” he added.

Alex Boas Endorses C2PA: New Digital Certificates

Bores' proposal is an open source standard called C2PA (Coalition for Content Provenance and Authenticity). This is a metadata standard that acts as a “tamper-proof” credential that is attached to files. You can record the origin of files. You can record the origin of files, whether they were taken with a physical camera or generated by AI. Editing evidence, such as how the media was modified and who created it. This is evidence of the person creating the media or the encrypted content.Boas said the challenge lies not in the technology, but in its implementation, as cryptography needs to become an industry standard for cryptographic proof to work. “The challenge is to get to the point where it becomes the default option because it requires the creator to attach an image,” Boas said, adding that the goal should be to get to the point where “if people see an image and there's no cryptographic evidence in that image, they should be suspicious.”He compared that trust to moving from HTTP to HTTPS. With the transition from HTTP to HTTPS, consumers instinctively know that they cannot trust banking sites that lack a secure connection. “It's like going to your bank's website and just loading the HTTP, right? You're instantly suspicious, but you can still generate an image,” he added.



Source link