Scrolling through Instagram Reels, Palo Alto High School senior Akio Altekar-Okazaki pauses on a video of a dog spraying a geyser of water out of its mouth. This video isn’t real. It was generated by OpenAI’s Sora 2, an artificial intelligence video generation tool and social media app.
Sora 2’s release on Oct. 1 prompted widespread discussion from major media outlets about the ethics of AI-generated videos and their effect on our society’s trust in the media. An NPR article cautioned that, while OpenAI promised guardrails, these measures don’t seem strict enough and misinformative content can still be freely generated.
Sora AI and other AI video generation tools crosses an ethical boundary. They blur the line between truth and fabrication, and threaten personal privacy and media credibility. If videos can be fabricated at the drop of a hat, they will lose their power as trustworthy records of our reality. Regulations are needed to guarantee that these platforms are not used for harm and misinformation.
Altekar-Okazaki said that he has begun to see a rise in Sora AI generated videos on social media, a trend he finds concerning.
“I’ve seen AI videos when I’m scrolling through my Instagram Reels,” Altekar-Okazaki said. “In the ones I’ve seen, it’s … being used as a joke, but it can be misinterpreted as reality. I’ve looked at the comments, and some people do seem to think that they [AI-generated videos] are real. I’ve seen political messages … that do seem kind of real, especially for the older population, they’re more susceptible.”
Altekar-Okazaki believes that advances in AI video generation have made it harder to tell real and generated videos apart.
“It’s kind of scary, because a few years ago, it was super easy to tell, but now I have to look for the watermark, … and there’s not always a watermark,” Altekar-Okazaki said.
This new advancement in technology may be revolutionary, but it raises serious moral questions. Is it ethical to use a tool that creates a fake reality? Verde’s answer is no — at least not without proper regulations. Although this tool can be used positively and allows users to generate videos using their own creativity, it also carries the risk of spreading misinformation and creating misleading, harmful content.
The dangers of unchecked AI-generated video content can already be observed in real political contexts. On Nov. 10, Republican Rep. Mike Collins’ team posted a 20 second AI video of Democratic Sen. Jon Ossoff on the platform X (formerly Twitter), with Ossoff saying that he voted to keep the government shut down because he didn’t care about farmers, and that he could never say “no” to Democratic Senate Minority Leader Chuck Schumer. The clip was viewed over 140,000 times on X.
Ossoff did not speak any of these words. The fake footage of Ossoff included a small and subtle water mark on the bottom right corner of the video labeled, “This video is AI-generated,” which could be easily missed or trimmed out. If this becomes a normalized practice, we may begin to see an increase in deceptive political campaigns that would make it difficult to trust what is on the internet.
According to Palo Alto High School history teacher Stephen Foug, AI generated clips like the Ossoff video may not even require viewing in order for their misinformation to spread.
“It might not even be people seeing the actual video,” Foug said. “Other people could see the video and just tell people they saw a video of something, and just the way rumors kind of get around like that.”
AI has no place in politics, and we should enforce regulations such as forcing AI companies to create large, visible watermarks for both audio and visuals.
Watermarks, which are often cropped out, shouldn’t be the only explicit indicator of AI content. The best way to limit the misinformation, distrust and high energy costs of these technologies is for the government to regulate AI companies and their tools before video loses its value as a record of reality.