How do I check if a news broadcast video is fake?

Learn how to verify viral news broadcasts and breaking video clips. We explain how to check video provenance, audio splicing, and AI-generated deepfakes.

Fact-checking · Published 2026-10-04 · 8 min read · 1,737 words

AI-generated illustration. A smartphone screen displaying a news video with a verification overlay showing a timeline of publication dates.

When a major international incident occurs, the first hours are crucial for gathering facts. They are also the exact window when fabricated videos do the most damage. On September 30, 2026, Flydubai flight FZ1073 from Dubai to Tel Aviv was forced to make an emergency landing in Saudi Arabia after an altercation resulted in a pilot being stabbed. This was a real, verified event. However, within hours, social media feeds were flooded with videos claiming to show the attack, the perpetrators, and the geopolitical fallout.

Almost all of the viral footage was fabricated. Opportunists and coordinated campaigns use real incidents to push their own agendas, as The Journal reported regarding the flight. Understanding how these videos are constructed is the first step in learning how to verify them yourself.

The anatomy of a fake broadcast

Misinformation rarely involves a completely AI-generated video from scratch. Generating a realistic, minutes-long news segment with perfect physics and lighting is still difficult. Instead, bad actors rely on manipulation, relabelling, and audio splicing.

The most prominent fake from the Flydubai incident was a video styled as a BBC News broadcast. The presenter in the video claimed that an Indian national named Raghav Menon attempted the hijacking as part of a joint RAW-Mossad intelligence operation to implicate Iran. As France 24 confirmed, this was an AI-generated deepfake. The BBC never broadcast the report. The creators likely took real footage of a newsreader and used AI tools to alter the lip movements and generate a synthetic voice track matching the false script.

A second viral clip claimed to show terrifying footage taken from a passenger window during the attack, featuring a plane descending rapidly while passengers screamed. Full Fact investigated and discovered the video was highly misleading. The visual footage showed a completely different flight passing over Dubai. The audio of the screaming passengers was stripped from a separate, older video of severe turbulence recorded in 2022.

A third video falsely claimed that NewsGuard reported the US Democratic Party orchestrated the attempted terrorist attack. NewsGuard confirmed they issued no such report, noting the clip appeared to be tied to a Russian disinformation campaign.

How to verify a news video manually

You do not always need complex software to debunk a viral claim. A few manual verification steps will expose most fabricated news clips.

First, check the primary source. If a major outlet like the BBC broadcasts a story about an international intelligence conspiracy, it will be the top headline on their official website. If you cannot find the matching article or video on the broadcaster's own platform, the clip circulating on social media is almost certainly a fake.

Second, separate the audio from the video. Because audio splicing is so common, listening to the track in isolation often reveals inconsistencies. Banging noises, screams, or sirens are frequently lifted from older disaster videos. You can take a screenshot of the video and run a reverse image search to find the earliest documented publication of the visual footage. If the footage appears in a 2022 travel vlog, it cannot possibly depict a 2026 hijacking.

How wyper checks claims and videos

When manual searching takes too long, you can use the free wyper web app or the browser extension to verify claims. The tool returns a truth score from 1 to 10 alongside an evidence chain of real, linked sources. Sources are never filtered by origin, meaning there is no automatic mainstream bonus and official sources are not inherently preferred. The evidence is presented so the reader can decide.

The standard check runs on a Dual-AI cross-check powered by Gemini and Grok. In this setup, Grok searches for its own sources, ensuring the two models do not share a single evidence pool.

If you need deeper analysis, the Full Spectrum option uses a Triple-AI reading setup where three models read one evidence set. The third voice does not search the web. It receives exactly the sources the main model received and reads them independently. A disagreement here is a different reading of the same material, not a different search result. We tested this extensively using 30 claims with frozen sources. Gemini and Grok agreed in 29 of 29 cases, even on deliberately contested claims. Two models trained on overlapping text are correlated, not independent, which means being confirmed by two AIs carries little information. However, the third model of different origin disagreed in 3 of 26 cases, every time providing a machine-verified quote. This independent reading argues against relying solely on correlated models, a claim we used to make ourselves.

The evidence chain itself is highly reliable. In our own published measurements of 633 real citations re-fetched, 73 percent were found verbatim on the page. Another 19 percent of pages were unreachable for automated readers due to paywalls, PDFs, or 403 errors, and only 6.4 percent were not findable. Importantly, the truth score is never proof of trustworthiness, and the tool never judges the intent of the poster. Both of those judgments stay entirely with you.

Verifying video context and provenance

Checking a video requires a different approach. The wyper extension uses subtitles when they are reachable. When subtitles are unavailable, an AI model actually watches the video within a specific time window. On the free tier, it watches up to 10 minutes on YouTube, while Pro users get up to 30 minutes. For videos on X, the limit is 2 minutes on the free tier and 7 minutes on Pro. Longer videos on X are refused outright, not partially checked. Because the web app cannot see your logged-in YouTube session, subtitles are often unavailable there. The extension runs in your browser and can reach more context. For videos on X, both tools take the exact same route.

For deep video analysis, paid plans include the AI Spotter. This tool costs 4 credits per video and provides a provenance report. It never outputs a percentage of how fake or "AI-ish" something looks, because that number is the exact harm we exist to prevent. Instead, it checks layers: content credentials verified against 55 official trust anchors, file metadata, reverse image search, earliest documented publication, and quote verification. For video, it takes still frames and searches each one separately. A hit on any single frame is a hit on the video.

You must understand one crucial rule about content credentials (C2PA). A content credential sits in its own box beside the picture data. Any repackaging drops it, including a lossless repackaging that leaves every pixel untouched. Social platforms repackage files upon upload. Therefore, on a video taken off a social feed, that signature layer will almost always find nothing. That is the normal case, not a suspicion, and the report says so. We advise checking the original file instead, where the signature can still be intact.

Because signatures are stripped by platforms, what carries the real weight for video verification is the earliest appearance plus the event-versus-post date gap. That is exactly what catches the most common real video deception: old footage relabelled as new. A pixel detector cannot see that context at all.

Comparing your verification options

Who needs which: Casual readers who want to verify a suspicious headline or a written claim should use manual checks or the web app for immediate answers without installing anything. Researchers, journalists, and users who frequently need to verify YouTube videos or check claims directly inside their social media feeds should use the extension, which can reach subtitles and watch video content within the active tab.

The short version

Misleading news broadcasts usually rely on spliced audio or AI-generated voices applied to stolen footage, rather than entirely generated video. You can verify these clips by checking the earliest publication date and reading the primary sources. Tools that provide an evidence chain help you confirm the context without relying on flawed pixel detectors.

Keep reading