Over the last few years, my concerns with AI have grown exponentially. It's not just about the laziness of people who want to stop reading and writing, but also the ability to saturate the internet with fake videos, which can not just confuse people but be used to outright lie to them, as well. Now that Sora 2 has dropped, and got an unprecedented million downloads in just a few days, the internet is also flooded with ways to remove the watermark on the video feature, taking away the one marker that informed viewers that they were watching an AI video. What could go wrong? This was inevitable, and yet it only took around 5 days for it to happen. Everyone predicted it, and it's infuriating that the people behind Sora did not take it seriously. “At launch, all outputs carry a visible watermark,” OpenAI rather naively said in a blog post. “All Sora videos also embed C2PA metadata—an industry-standard signature—and we maintain internal reverse-image and audio search tools that can trace videos back to Sora with high accuracy, building on successful systems from ChatGPT image generation and Sora 1.”I'm not sure how they didn't anticipate people bypassing it that quickly, but it shows a complete lack of self-awareness or caring. There are already many TikTok videos going viral without a watermark, and now we have to worry about how easy it will be to bypass the image and likeness features they already promised, or all the copyright issues. -...
Published By: NoFilmSchool - Yesterday