Some current and previous Openai researchers have said about their first foray into social media: SORA app, Tiktok-style feeds filled with videos generated by AI, and many Sam Altman Deepfakes. Researchers who air their complaints on X seem to have been torn apart about how launch fits Openai’s non-profit mission to develop advanced AI that benefits humanity.
“AI-based feeds are scary,” Openai pre-registered researcher John Hallman said in an X’s post.
“We share a similar combination of worry and excitement. Sora2 is technically surprising, but it’s premature to celebrate avoiding other social media apps and the pitfalls of deepfakes,” said Boaz Barak, another Openai researcher and Harvard professor.
Former Openai researcher Rohan Pandey used the moment to block the regular lab of a new startup, consisting of former AI lab researchers looking to build AI systems for scientific discovery.
There were many other posts along the same line.
The SORA launch highlights Openai’s core tensions that burn over and over again. It is the fastest growing consumer technology company on the planet, but it is also the Frontier AI Lab, which has a noble non-profit charter. Some former Openai employees I discussed spoke to claim that in theory the consumer business would be useful in the mission. ChatGPT helps fund AI research and distribute technology widely.
Openai CEO Sam Altman said in a post on Wednesday in X that he is addressing why the company allocates so much capital and computing power to AI social media apps.
TechCrunch Events
San Francisco
|
October 27th-29th, 2025
“We primarily need capital for build AI that can do science. Certainly, almost every research effort focuses on AGI,” says Altman. “It’s also great that along the way people make new tech/products cool and smile, and that shows that they make money considering all the needs.”
“When we launched ChatGpt, there were a lot of ‘I needed this and where are I?'” continued Altman. “Eality is subtle when it comes to optimal trajectories for (r) companies.”
But at what point will Openai’s consumer business overtake its non-profit mission? In other words, Openai is at odds with missions, so when do you say no to the opportunity to make money and develop a platform?
This question scrutinizes the looming transition of Openai’s for-profit organizations as regulators. This must be completed to raise additional capital and ultimately make it public. California Attorney General Rob Bonta said last month that he was “particularly interested in ensuring that Open I safety missions as a nonprofit remained as front and center.”
Skeptics have rejected Openai’s mission from Big Tech as a branding tool to seduce talent. However, many Openai insiders claim that they are central to why they joined the company in the first place.
Sora’s footprints are small for now. The app is one day ago. But its debut marks a massive expansion of Openai’s consumer business, exposing the company to incentives that have plagued social media apps for decades.
Unlike ChatGpt, which is optimized for usefulness, Openai says Sora is built for fun, a place to generate and share AI clips. The feed feels closer to Tiktok or Instagram reels, a platform notorious for addictive loops.
Openai argued that they wanted to avoid these pitfalls, claiming in a blog post announcing the launch of SORA, “Dookscrolling, Addiction, Isolation and RL-Sloptimized Feeds are heartfelt.” The company explicitly states that it is not optimized for the time spent on feed and instead wants to maximize creation. Openai says it sends reminders when users are scrolling too much, and mostly shows people they know.
This is a more powerful starting point than the atmosphere of meta (another AI-powered short video feed released last week). As a former Openai policy leader, Miles Brundage points out that there may be good and bad applications for AI-Video feeds, as we saw in the chatbot era.
Still, as Altman has long admitted, no one sets up to build an addictive app. Incentives to run feeds guide them. Openai is encountering issues with ChatGpt’s psychofancy. This is not intentional, the company says, due to some of the training techniques.
On a June podcast, Altman discussed what he calls “a big inconsistency on social media.”
“One of the big mistakes in the social media era was that feed algorithms have unintended negative consequences across society. And perhaps even individual users were doing what users wanted, but at this point, they could continue to spend their time on the site.”
It’s instantly to tell you how well the SORA app is tuned with your users or Openai missions. Users have already noticed technologies that optimize app engagement, such as the dynamic emojis that appear whenever they like the video. It feels like it’s designed to shoot a little dopamine at users to participate in the video.
The actual test is how Openai evolves Sora. Given how much AI has taken over its regular social media feeds, it seems unlikely that AI-Native feeds can have moments anytime soon. It remains to be seen whether Openai can raise Sora without replicating the mistakes of its predecessor.
