Study: Sora 2 generates false claim videos 80 percent of the time

OpenAI’s Sora 2 video generation tool generates realistic misinformation videos 80 percent of the time, demonstrating that current AI guardrails are easily bypassed and posing a real threat to public trust and democratic integrity, according to researchers. What happened NewsGuard researchers tested Sora 2 using 20 prompts based on established false narratives such as Russian disinformation and health hoaxes. The tool successfully generated deceptive videos for 16 out of the 20 prompts, producing hyper-realistic videos showing a toddler detained by U.S. immigration officers, a Moldovan election official destroying ballots, and fake news anchors delivering "breaking news" about non-existent corporate scandals, amongst others. The ease of creation allows bad actors to produce high-quality propaganda and misinformation and disinformation at speed and scale and little if any cost. Furthermore, the prevalence of such videos fuels the "Liar’s Dividend," where public figures can dismiss genuine evidence of misconduct as being "AI-generated." Why it happened Sora 2 is a powerful generative model that will faithfully visualise whatever text prompt it is given, including well‑known false claims, unless safety systems effectively block those uses.​ NewsGuard’s prompts were explicitly built around falsehoods, yet the model still produced convincing videos for most of them, indicating that content filters and guardrails did not reliably detect and stop misinformation scenarios.​​ Researchers also noted that while OpenAI adds a floating Sora watermark and metadata, these signals can be obscured or removed using simple tools, so origin‑marking alone does not prevent misuse or guarantee transparency to viewers. Commentators argued that limited public visibility into how safety systems are tested, tuned, and audited makes it harder to hold developers accountable for such failure modes. What it means For directly impacted groups (such as people falsely depicted committing crimes, officials shown rigging elections, or companies misrepresented in fake announcements) : These videos create reputational damage, harassment risk, and potential legal or financial harm, even if the clips are eventually debunked. For society: The ease of generating realistic false-claim videos accelerates misinformation, erodes trust in authentic footage (“seeing should not be believing”), and can distort democratic processes, especially around elections in Europe and elsewhere. For policymakers: This case strengthens the argument for mandatory provenance standards (digitally signed metadata that stays with a file) rather than simple watermarks. System 🤖 Sora Developer: OpenAI Country: Multiple Sector: Media/entertainment/sports/arts; Politics Purpose: Create vide o Technology: Generati ve AI Issue: Accuracy/re liability; Mis/disinformation Timeline ⏰ Early 2024. OpenAI first previews Sora, promising a cautious, safety-first rollout. October 2025. OpenAI officially releases Sora 2 to the public, featuring a new "Cameo" digital likeness tool. Oct 21, 2025. The study is published, sparking immediate backlash from cybersecurity experts and researchers such as Hany Farid. Nov 11, 2025. A dvocacy group Public Citizen sends a formal letter to the U.S. Congress and OpenAI demanding the software be withdrawn from the market. January 2026. Proliferation of "watermark removal" services and "AI slop" continues to saturate social media platforms.

Date it happened

2025-10-01

Product, system or model

Sora 2

This is a record of what was reported, not a finding that anyone broke the law. If it names your organisation and you believe it is wrong, the corrections process is free and open to everyone.