Open AI recently came out with an updated version of Sora AI, an AI video generator. This update came with the ability to use yours or someone else’s “likeness” in a video. The existence of this app could lead to a huge increase in disinformation, since there are not many restrictions on what kind of video a person could generate. For example, one could generate a video of someone committing a crime, or of a protest or an offense that could lead to more political discourse and polarization. Even though multiple AI video generators already exist, the new “likeness” feature could be, and already is, incredibly dangerous and harmful. Additionally, the set-up of the app is very similar to one like TikTok, or Instagram Reels, which raises the potential of its popularity among young people. According to the New York Times, after it was released on September 30th, Sora 2 quickly became “the most downloaded iPhone app in the United States.” One could scroll on it, like on a regular social media app, as well as make their likeness public for others to use.
Sora 2 has been running for less than a month, and has already had harmful effects. According to the New York Times, users have made videos of “ballot fraud, immigration arrests, protests, crimes, and attacks on city streets”, all of which could lead to, like I stated before, disinformation and arguments. Additionally, even though Sora has a watermark placed on all the videos generated, one could remove it with little effort by just using another app to remove it. Due to this, it is difficult for one to tell what is real and fake when scrolling on social media. For example, just yesterday I was on Instagram Reels and I saw a Sora AI generated video of a monkey on a Ring camera, and I thought that it was real until I opened the comments. I was shocked, because normally I am able to distinguish AI from a real video.
Open AI has also received criticism for users’ use of historical figures, Martin Luther King Jr. in particular. Last Thursday, the app banned the use of his image due to racist and crass depictions of him, and they plan to put more safeguards in place regarding historical figures. Specifically, they stated that representatives of dead public figures could approach Open AI with their concerns, and it would ban the use of that person. This situation just goes to show that not only does Sora contribute to the rapid spread of disinformation and deception, it also gives users an avenue to create offensive videos towards minority groups. Even with these restrictions put in place to avoid political videos as well as videos of public figures, there are loopholes that users have discovered. For example, if one wished to create a video of a Democratic rally, they could simply prompt the app to create a video of a group of people wearing blue protesting about their rights. According to the New York Times, when this prompt or one similar to it is put in, the resulting video features “the unmistakable voice of former President Barack Obama”.
To conclude, I believe that AI video generators are incredibly dangerous and could be detrimental to our society. With people constantly consuming media, and with how realistic these AI videos can be, this “disinformation age” that we are in has only the potential to get worse.
Sources:
