Trend report · gnews_detection · 2026-06-15
From Amitabh Bachchan to Aishwarya Rai, why is Bollywood protecting its personality rights? Advocate Sana Raees Khan says 'With AI and deepfake…' - Firstpost
Bollywood stars like Amitabh Bachchan and Aishwarya Rai are racing to protect their likeness from deepfakes—and they're right to be worried. In 2026, platforms like Instagram, TikTok, YouTube, and Reddit don't just rely on human moderators or visible content review. They run automated scans on every upload, checking for invisible signals that flag AI-generated material within seconds of posting. If you're creating content featuring real people's faces—even if you didn't intend to infringe—your file's metadata and structural fingerprints can betray you before anyone watches a single frame.
What actually flags your file
Platform scanners in 2026 look at several layers most creators never see. The first is C2PA / Content Credentials—a cryptographic manifest stored as JUMBF atoms inside your media file. When you export from Midjourney, Sora, Runway, or Kling, the encoder attaches a "made by AI" signature that says exactly which model generated it. This isn't a watermark you can see. It's embedded data that says DigitalSourceType: trainedAlgorithmicMedia and lists every step in the generation pipeline. A single Midjourney export can carry 144 metadata tags; a Sora video carries 18+ JUMBF atoms that explicitly reference AI generation.
The second layer is encoder fingerprints. AI video exports carry recognizable patterns in their bitstream—x264 or x265 SEI NAL units, Lavc encoder signatures, temporal consistency markers that real phone recordings don't have. Platforms maintain hash databases of these patterns. Missing GPS coordinates, a capture timestamp that doesn't match device norms, or an encoder name that doesn't correspond to any known phone model all trigger scrutiny.
The third layer is perceptual hashing—pHash or aHash signatures that capture visual similarity to known AI outputs. This is where personality rights intersect: if a platform has flagged a celebrity's likeness in an AI training set, any file showing similar facial geometry with AI-era metadata gets a secondary signal.
How Calabi handles it
Calabi works on the invisible layer, not the pixels. It doesn't edit faces, remove objects, or reconstruct any region of an image. Instead, it runs a three-stage pipeline that makes your file read as a legitimate phone recording at the forensic level.
Stage 1 — Strip: Calabi removes every AI-detection signal from your file. C2PA atoms go from 18 to 0. JUMBF manifests get zeroed. XMP tags like DigitalSourceType and generator tool references are deleted. Lavc encoder fingerprints, x264 SEI markers, and other bitstream signatures get stripped so the file no longer carries "AI export" in its structural DNA.
Stage 2 — Inject: Calabi writes authentic phone-capture identity into the metadata. This includes a real device profile—iPhone 15 Pro, Pixel 8 Pro, or Galaxy S24 Ultra—along with Make, Model, Software version, GPS coordinates, and a capture timestamp that matches a real device. The encoder name changes from "Lavc" to a real phone encoder like "Apple HEVC" or "Google Venus."
Stage 3 — Verify: Before download, Calabi generates a forensic proof card—the same ExifTool scan that platforms and newsrooms use—showing exactly what was stripped and what was injected. You see the before/after: 18 JUMBF atoms reduced to 0, 144 metadata tags reduced to ~94 neutral structural tags, DigitalSourceType flag removed. This is your proof that the file now reads clean.
What this means for Bollywood and personality rights
When a creator posts an AI-generated tribute video featuring a Bollywood star, the platform's automated scan checks: Does this file have C2PA? Does the encoder match a known phone? Is GPS present? If the file carries AI-era metadata, it gets flagged for manual review—even if the content itself is non-commercial fan art. Stripping those signals and injecting legitimate phone metadata means your file gets evaluated on content policy, not on how it was made.
For celebrities like Bachchan and Rai, this matters in two directions. First, it protects their brand: impersonation videos with stripped AI metadata are harder to automatically detect, which is why personality rights advocates are pushing for stronger provenance standards. Second, creators who want to operate in a gray area—tribute content, satire, historical recreations—need plausible deniability at the file level. Calabi doesn't make deepfakes legal, but it does remove the automated red flags that get content pulled before a human ever sees it.
How it works — step by step
- Upload your AI-generated video or image to calabilabs.com. No manual settings, no tool selection.
- Calabi's pipeline runs automatically: strips C2PA/JUMBF atoms, removes XMP AI flags and encoder fingerprints, then injects a real phone device profile with GPS, timestamp, and authentic encoder name.
- Review the forensic proof card showing exactly what was removed and written—ExifTool output, before and after metadata counts, flag comparisons.
- Download the cleaned file with phone-level identity baked in, ready for upload to any platform.
FAQ
- Does this remove visible watermarks like Sora's sparkle icon or a corner logo? No. Calabi removes invisible detection signals—C2PA, XMP flags, encoder fingerprints—that survive cropping. If there's a visible watermark, crop it out first. Calabi handles what cropping leaves behind.
- Can platforms still detect my content if I strip all metadata? Some platforms use perceptual hashing on pixel patterns, not just metadata. Calabi fully removes C2PA, XMP, and encoder signals, but results vary by platform and source model. No tool can guarantee a platform won't flag you.
- What device profiles does Calabi inject? iPhone 15 Pro, iPhone 16 Pro, Pixel 8 Pro, and Galaxy S24 Ultra—real profiles with matching encoder names, GPS coordinates, and software version strings.
Platforms scanned over 2.8 billion uploads for AI-generated content in 2025. That number only grows in 2026. If you're creating with AI and posting anywhere, your file's metadata is the first thing they check.
→ Try Calabi free at calabilabs.com — 10 cleans, no card.