The IPTC has launched Metawatch, a monthly monitoring tool that tracks whether news publishers retain embedded photo metadata during the publishing process. The data indicates that most publishers strip critical metadata, which complicates provenance and rights management as AI crawlers increasingly scrape web content.
The widespread removal of embedded metadata creates a critical vulnerability as AI companies like OpenAI and Google increasingly crawl web content for training data. While 46% of publishers now block crawlers via robots.txt, that protection fails once an image is syndicated or shared; the embedded file data remains the only persistent record of ownership and usage rights. For the streaming and digital media ecosystem, this technical gap complicates the adoption of C2PA Content Credentials and generative AI disclosures. Industry professionals should monitor if monthly Metawatch reporting incentivizes CMS and CDN vendors to update default settings that currently prioritize minor byte savings over rights management.
The IPTC has been tracking metadata stripping for nearly a decade. Its earlier Photo Metadata Survey, conducted biennially from 2016 through 2024, consistently found that between 85% and 92% of images published by major news outlets lost their embedded IPTC fields during CMS processing. Metawatch represents a shift from periodic surveying to continuous monthly monitoring, giving rights holders and standards bodies a real-time dashboard rather than a retrospective snapshot. The tool's launch coincides with growing pressure from the Coalition for Content Provenance and Authenticity (C2PA), whose Content Credentials specification relies on embedded metadata as a binding layer for provenance claims. When publishers strip that metadata at ingestion, C2PA manifests become orphaned from the pixel data they are meant to protect.
On the business and regulatory side, the European Union's AI Act, which entered into force in August 2024 and begins phased enforcement in 2025 and 2026, requires AI providers to disclose training data sources and respect opt-out mechanisms. The Act's transparency obligations for general-purpose AI models explicitly reference machine-readable rights reservations, a category that includes both robots.txt directives and embedded metadata such as IPTC fields. This creates a legal incentive for publishers to retain metadata: if an image carries a rights reservation in its IPTC block and an AI company ignores it, the publisher has a clearer evidentiary trail. Conversely, stripped metadata weakens the publisher's ability to prove that a reservation existed at the time of crawling.
Adobe and Google, both named in the IPTC announcement as companies whose AI systems scrape web content, have taken divergent technical approaches to provenance. Adobe co-founded C2PA in 2021 and has integrated Content Credentials into Photoshop, Lightroom, and Firefly, embedding signed provenance manifests at the point of creation. Google, meanwhile, has focused on SynthID watermarking for AI-generated images and has not adopted C2PA manifests in its core products. The IPTC has noted that C2PA credentials and IPTC metadata are complementary rather than competing standards, with C2PA providing tamper-evident provenance chains and IPTC providing human-readable rights and descriptive fields. Metawatch data will likely inform whether metadata automation for broadcasters vendors update default export settings to preserve both layers simultaneously, a change that would benefit the entire content supply chain from photographers to AI training-data auditors.
The IPTC has launched Metawatch, a monitoring tool tracking how 500 global news publishers handle image metadata. Initial findings reveal 90% of images lose critical rights data during the publishing process. This matters because stripped metadata leaves content vulnerable to unauthorized AI scraping and complicates the adoption of C2PA provenance standards.
Metawatch is a new monitoring tool from the IPTC that provides continuous monthly tracking of how global news publishers handle image metadata, replacing previous biennial surveys.
Initial Metawatch data indicates that 90% of sampled images lose their embedded IPTC metadata, with only 11% of images retaining any information.
Embedded metadata serves as a persistent record of ownership and rights. When publishers strip this data, it becomes difficult to prove rights reservations to AI companies that scrape web content for training data.
The '4 Cs' refer to creator, caption, copyright, and credit line, which are critical fields that three out of four publishers currently fail to retain.
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source