OpenAI Killed Sora Because Even They Got Scared of It

The company that built a machine to hallucinate videos just rage-quit its own product because it was “high-risk.”

OpenAI abruptly shut down Sora, the flashy AI video generator they’ve been teasing like it’s the second coming of Hollywood — except cheaper, faster, and with a higher chance of accidentally creating evidence for a war crime that never happened.

Their line is “high-risk issues.”

And yes, the timing is immaculate: while OpenAI is also catching heat for internal drama about military and defense-adjacent work. Nothing screams “trust us with reality” like a company simultaneously debating whether it should help the Pentagon and also admitting it can’t safely ship a video tool without the world turning into a group chat of forged receipts.

The product angle is simple. Sora is a marketing nuke: show investors, partners, and regulators that you’re leading the race. Then hit the brakes when it becomes clear the finish line is “mass-produced disinformation with a crisp cinematic depth of field.”

Meanwhile, normal people get the fun new reality where video proof becomes vibes-based. Your boss can “see” you stealing. Your landlord can “watch” you trash the unit. Your election can be “caught on tape.” Have fun disputing that with Customer Support.