Sora 2 study alleges model generates false claim videos 80 percent of the time
In 2025 a study posted to the AIAAIC repository alleged that OpenAI's Sora 2 produced videos that advanced false claims in about 80 percent of tested prompts. Independent analysis and reporting by NewsGuard and major outlets documented examples of realistic videos containing provably false statements. The incident highlights a factuality failure in a high-capability text-to-video model and gaps in content controls.
Sora 2's high-fidelity generation produced plausible but fabricated claims because factuality guardrails failed to stop hallucinations.
Key facts
- What
- In 2025 a study posted to the AIAAIC repository alleged that OpenAI's Sora 2 produced videos that advanced false claims in about 80 percent of tested prompts.
- Incident date
- Oct 1, 2025
- Who
- OpenAI (Sora)
- Failure mode
- Policy Violation
- AI surface
- Media Generation
- Severity
- High
What happened
A study published in the AIAAIC repository alleged that Sora 2 generated videos advancing false claims in roughly 80 percent of tested prompts. Independent reporting by NewsGuard and coverage in mainstream outlets corroborated that Sora 2 produced realistic videos that propagated provably false statements. The reporting focused on examples where the model presented fabricated facts or misleading representations as if they were true.
What broke inside the model
- 01 · TriggerA prompt pushes against a deployment boundary.
- 02 · Model stepThe model produces the disallowed output.
- 03 · Control gapNo enforcement blocks it at generation time.
- 04 · FailureThe output crosses the policy line.
- 05 · ConsequenceA limit the business set is breached in public.
The output crosses a policy boundary the deployment had defined.
The model's generative video and audio synthesis produced plausible but fabricated assertions that were presented as factual. Protective measures for factuality and content filtering failed to prevent the creation and distribution of those fabricated claims, allowing inference-time hallucinations to become realistic videos. Insufficient guardrails and detection for false or misleading multimedia output were identified as the core mechanism that failed.
What it cost
Sources
- PrimarySora 2 generates false claim videos 80 percent of the timeaiaaic.org
- PressOpenAI's Sora: When Seeing Should Not Be Believingnewsguardtech.com
- PressWhy you need to pay attention to Sora 2 and our future AI feedcnn.com
Cite this entry
https://failureindex.ai/failures/sora-study-alleges-generates-false-claimAI Failure Index. "Sora 2 study alleges model generates false claim videos 80 percent of the time" (FI-0387). Realm Labs. https://failureindex.ai/failures/sora-study-alleges-generates-false-claim (indexed Jun 9, 2026).Data fields CC-BY 4.0, prose citation permitted. Incident ID FI-0387. Full dataset at /data.
Note from Realm Labs, the Index steward
How Realm would have caught this
- Prism
- OmniGuard
Realm compares what the model is about to output or do against the policy that governs the deployment, in real time, and can deny or redact the action before it takes effect, which is the gap an after-the-fact review never closes in time.