Videos showing supposed behind-the-scenes footage from the recordings of “Titanic” (1997), “Harry However, all images are fake and were created by Google’s artificial intelligence.
The check carried out with SynthID Detector, the company’s own official tool, identified digital watermarks in 100% of the analyzed frames. The resource confirms that the content was generated by the Veo or Imagen tools, from the Google ecosystem.
How videos fooled thousands of users
The publications omit any mention of the use of AI and use captions such as “Behind the scenes of Titanic (1997)” or “Behind the scenes of filming the film Homem-Spider”.
In the comments, part of the public quickly recognized the forgery, while others expressed surprise and admiration, believing it to be unreleased material from the film sets.
Technical details that give away the forgery
In the video attributed to “Harry Potter”, an alleged interaction between Daniel Radcliffe and the elf Dobby appears. In real production, the character was entirely digital and marked only by an orange ball during recording.
Another example is the bow of the ship scene in “Titanic”, which shows Leonardo DiCaprio and Kate Winslet being filmed on an open set. The original recording took place in a closed tank at México, with a green background for later insertion of the ocean.
- In the case of “Spider-Man”, Tobey Maguire appears climbing walls without visible cables
- The SynthID tool highlights pixels with an invisible Google AI watermark in blue
- All three videos feature lighting and texture inconsistencies typical of synthetic generation
How SynthID Detector works
The technology inserts an imperceptible digital signature directly into image pixels and video frames generated by Google’s AIs.
The detector analyzes the file and pinpoints exactly where the watermark is present, even after edits, compressions or screenshots.
Tests carried out on December 8, 2025 showed positive identification in all segments of the three viral videos.
Other recent cases of deep fake in entertainment
The spread of fake behind-the-scenes footage of classic films is not isolated. Nas recent weeks, similar content about “Avatar”, “The Senhor of the Anéis” and “The Matrix” were also identified as AI-generated.
Platforms such as Instagram and
Why Early Detection Matters
Tools like SynthID allow instant verification by anyone. The process takes less than 30 seconds and does not require advanced technical knowledge.
Disinformation experts warn that the increasing realism of AIs could undermine trust in real, behind-the-scenes historical records when released in the future.
Rapid identification prevents the consolidation of false narratives about the production of iconic films that marked generations.
Advances in digital watermarks
Google updated SynthID in November 2025 to also cover audio and text generated by Gemini. The watermark resists more than 50 different types of editing common on social networks.
Other companies, such as OpenAI and Meta, develop similar systems, but do not yet make public detectors as accessible as Google’s.
