A camera credential is only useful if the editing software carries it properly
Adobe's August 2026 Camera Raw 18.5 fixed-issues list includes a C2PA problem affecting Google Pixel 10 files. Adobe lists the issue among problems fixed in the release. The update is a practical reminder that capture-side provenance is only one part of an authenticity chain: credentials also have to be interpreted and preserved correctly by the applications photographers use afterwards.
Why visual creators should care
Professional provenance is increasingly being sold as a camera feature, but photographers rarely deliver untouched camera files. They pass through raw processors, DAM systems, editors, exports and publishing platforms. A failure at any one of those stages can weaken the evidence attached at capture.
Practical takeaway
If Content Credentials matter to a project, test a real file through the complete production path after software updates. Verify the camera original and the exported deliverable separately rather than assuming that a credential visible at capture survived processing.
What to watch next
Watch for further camera-specific C2PA compatibility fixes and whether Adobe provides clearer diagnostics when a credential is lost, changed or cannot be validated.
Sources
Midjourney wants Hollywood's own AI use pulled into the copyright case
A hearing scheduled for August 17 concerns Midjourney's challenge to limits placed on discovery in the copyright litigation brought by Disney, Universal and Warner Bros. Discovery. Midjourney has sought wider disclosure of the studios' internal AI use, including material beyond consumer-facing AI products. A magistrate judge previously restricted much of that discovery, and Midjourney has asked the district court to reconsider the limitation.
Why visual creators should care
The underlying infringement claims remain about Midjourney, but the discovery fight shows that internal AI practices can themselves become evidence once a company enters litigation over generative technology. Studios, agencies and other rights holders therefore need AI records not only to manage production but potentially to explain their own conduct in court.
Practical takeaway
Keep internal AI policies consistent with the position you take with suppliers and clients. Record which systems are permitted, what assets may enter them, and whether experimental or internal use differs from public-facing production.
What to watch next
Watch the court's ruling on the discovery dispute and whether the studios are required to disclose substantially more about internal AI systems, training, research or creative use.
Sources
China is treating missing AI labels as a platform compliance problem
China's Cyberspace Administration warned ByteDance products Jianying, Maoxiang and Jimeng AI in April over compliance with the country's rules for labelling AI-generated material. Jianying is a major video-editing product and Jimeng AI generates visual media, making the warning directly relevant to creator-facing image and video workflows.
Why visual creators should care
Labelling rules only work if platforms actually carry them through production and distribution. China's approach places compliance pressure on the service itself rather than relying only on the individual creator to remember a disclosure at publication.
Practical takeaway
Creators distributing synthetic visual work into regulated markets should not remove platform-generated labels or metadata during editing. Keep the original output and the final published version so you can show what disclosure was present at each stage.
What to watch next
Watch whether Chinese regulators move from warnings to penalties and whether editing applications begin preventing users from exporting generated material without the required label.
Sources
South Korea makes realistic generative output a transparency issue
South Korea's AI Basic Act and its enforcement framework took effect on January 22, 2026. Among its transparency requirements, providers must notify users about generative AI and clearly label generated output where it may be difficult to distinguish from reality. Authorities have provided a grace period before administrative fines are imposed.
Why visual creators should care
Photorealistic output sits at the centre of the rule. For image and video providers serving Korean users, disclosure is therefore becoming part of product design rather than something left entirely to agencies or creators downstream.
Practical takeaway
When using a visual generator for international client work, record which jurisdiction the finished media will be distributed in. A compliant generation workflow in one market may need additional labelling or notification elsewhere.
What to watch next
Watch the transparency guidance issued during the grace period and how Korean regulators distinguish obvious creative imagery from generated photos or videos that could reasonably be mistaken for real capture.
Sources
AI video is starting to replace the shoot rather than assist it
Fresh reporting on China's short-form microdrama industry describes producers shifting from conventional actors, crews and rapid studio shoots towards AI-generated video following advances in systems such as Seedance 2.0. PetaPixel reports claims of substantial cost and production-time savings alongside sharp declines in work and income for some conventional production businesses and performers.
Why visual creators should care
This is a clearer example of generative video competing with an entire production chain rather than replacing a single stock clip or post-production task. The commercial question for filmmakers is moving from whether AI belongs inside the workflow to whether a physical shoot happens at all.
Practical takeaway
Production companies should make the production method explicit in briefs and estimates. If a client expects performers, cinematography or real product footage, define those requirements rather than assuming the word 'video' implies physical capture.
What to watch next
Watch whether clients begin requesting camera-origin guarantees for particular categories of commercial video and whether performers or crew agreements address synthetic replacements and AI-generated versions of established productions.
Sources
Fully generated video breaks an old assumption about fake visual evidence
New research on text-to-video fake news argues that modern generators create a different forensic problem from older manipulated footage. Traditional fake-video systems often look for mismatches between a claim and recycled or edited source footage. A text-to-video system can instead generate a nonexistent event that visually matches the false claim perfectly. The researchers created a dataset separating real video, conventional 'cheap fakes' and fully synthetic fake-news footage and report improved detection when semantic reasoning is combined with lower-level generation traces.
Why visual creators should care
For journalists, documentary filmmakers and publishers, visual consistency is no longer reassuring. A completely fabricated clip can look internally coherent precisely because every element was generated to support the same invented story.
Practical takeaway
Treat verification as an origin problem as well as an image-forensics problem. For evidential video, ask where the footage came from, who captured it and what original file exists rather than relying mainly on whether the pixels look believable.
What to watch next
Watch independent testing against current commercial video models and whether provenance signals outperform classifiers once synthetic footage has been compressed, edited and reposted.
Sources
