Calabi Labs · Guide · 2026-06-12
Meta changes made with ai policy after mislabeling images pcmag
Meta Changes "Made With AI" Policy After Mislabeling Images: The Full Story
Meta has updated its AI labeling policy after widespread complaints that the company was incorrectly tagging real, human-made photographs as "Made With AI" on Facebook, Instagram, and Threads. The change, which PCMag and other outlets documented in July 2024, came after photographers discovered their authentic work was being flagged as AI-generated—simply because they'd used standard editing tools.
What Happened
In early 2024, Meta began rolling out "Made with AI" labels across its platforms. The goal was transparency: let users know when content was created or significantly altered using generative AI. But the detection system had a flaw.
Meta's labeling relied on two main methods:
- Metadata signals from image files (C2PA, IPTC standards that flag AI-generated content)
- AI-similarity detection that compares uploaded images against known AI-generated patterns
The problem? Both methods were catching legitimate photos. Photographers who used Adobe Photoshop's AI-powered tools for routine edits—like noise reduction, subject-aware fill, or color grading—were having their work automatically labeled "Made with AI" even though the final image was still a photograph they'd taken.
The backlash was immediate. Professional photographers called it misleading and damaging to their work's integrity. A photograph that's been color-corrected isn't "made with AI"—it's a photograph that's been edited.
What Meta Changed
Meta acknowledged the mislabeling issue and revised its approach. The key changes:
- Narrower criteria for AI labels: Meta tightened the definition of what qualifies as "Made with AI" content. Standard edits made with traditional tools (even AI-assisted tools within Photoshop or Lightroom) no longer automatically trigger the label unless the AI generation was the core creative process.
- Clearer distinction between generated and modified content: Meta shifted language to be more precise—distinguishing between content that was AI-generated (created from scratch by AI) versus content that was AI-modified (a real photo touched by AI tools).
- User controls for disputed labels: Meta introduced a way for users to dispute or request removal of the "Made with AI" label if they believe it was applied incorrectly.
- Less aggressive metadata detection: Rather than automatically flagging any image with AI-tool metadata, Meta moved toward a more human-reviewed approach for contested cases.
Why This Matters
This incident highlighted a broader problem in the AI content detection space: metadata can be spoofed or added unintentionally, and similarity detection is imprecise. Photographers, illustrators, and digital artists have been caught in the crossfire of automated AI labeling systems across multiple platforms.
Meta's policy adjustment signals that platforms are learning: overly broad AI detection causes real harm to real creators, and policy needs to account for the nuance of how humans actually use AI tools.
The Bigger Picture
The mislabeling controversy is part of a larger industry reckoning. As AI-generated content proliferates, platforms face pressure to label it clearly—but the technical tools for detection remain unreliable. Metadata is easily stripped, easily added, and often present in images that contain zero AI generation. Similarity models are improving but still flag human work as AI at unacceptable rates.
Meta's course correction suggests the company is prioritizing creator trust over aggressive AI labeling—a notable shift given how heavily platforms have leaned into AI features.
Try Calabi free at calabilabs.com — 10 cleans, no card.