A photographer can now call Photoshop and Lightroom from inside an AI conversation
Adobe announced on September 24 that Photoshop, Lightroom, Adobe Express and Firefly capabilities are now available directly through Google Gemini. A user can upload product photographs, describe the required result and have Adobe tools handle operations such as lighting and colour optimisation, consistent cropping and preparation of images for storefronts or marketing. Adobe has also expanded its Claude plugin to more than 80 tools across applications including Photoshop, Premiere, Lightroom, Illustrator, InDesign, Express, Adobe Stock and Acrobat. A new interactive Express editor allows individual images, copy, colours and fonts to be changed without regenerating the whole design.
Why visual creators should care
The application named in a project record may no longer tell you where the creative decision was actually made. A photograph can enter Gemini, be altered through Adobe technology and continue into another publishing workflow without the photographer ever beginning inside Lightroom or Photoshop.
Practical takeaway
Record the operation and underlying creative service, not just the conversational interface. If Gemini invokes Adobe to alter a client photograph, 'Gemini' alone is an incomplete production record.
What to watch next
Watch how Adobe exposes Content Credentials, Firefly-generation information and edit history when its tools are invoked from third-party AI assistants rather than from Adobe's own interfaces.
Sources
Austria is putting Content Credentials into the publishing pipeline rather than adding them afterwards
The Austrian Parliament is introducing C2PA Content Credentials across newly produced video-on-demand material. Broadcast technology company Big Blue Marble is adding the credentials automatically through its Cloud Video Kit as parliamentary video is processed for publication. The resulting credential can contain information about source, editing history and a digital signature that compatible services can inspect. The system does not stop somebody copying or subsequently manipulating the video; it provides a verifiable record of the signed official version.
Why visual creators should care
This is a useful real-world model for newsrooms, agencies and production studios. Provenance becomes much more dependable operationally when it is part of the automated publishing pipeline rather than an optional task somebody has to remember before delivery.
Practical takeaway
Where authenticity matters, automate provenance at export or publication. A workflow-level credential is harder to forget than a manual provenance step performed asset by asset.
What to watch next
Watch whether broadcasters, newsrooms and government communication teams adopt automatic C2PA signing and whether social and video platforms preserve those credentials after transcoding.
Sources
Putting somebody's photograph into AI video creates more than one permission
VerseIn's revised terms took effect September 24. The service creates AI images and videos, including placing a person from an uploaded photograph into video. VerseIn says users retain their rights in inputs and may use generated outputs, but describes that output right as contractual permission rather than a guarantee of copyright, originality or third-party clearance. It says inputs and outputs are not used to train AI models and that its agreements require AI providers to do the same. For photographs of people, users may upload only themselves or people who have consented. VerseIn says the process extracts facial data, which may constitute biometric information, and requires users uploading another person to have whatever consent or written release applicable law requires.
Why visual creators should care
The source photograph, the person's likeness, facial data and resulting synthetic video can each involve different rights. A model release written for an ordinary portrait session may not answer all four questions.
Practical takeaway
For synthetic people, record permission to use the source photograph, permission to create the new depiction and any required biometric-processing consent separately. Do not rely on the platform's output permission as proof that all likeness rights are cleared.
What to watch next
Watch whether professional AI-video services standardise portable consent records that identify the source photograph, depicted person and authorised synthetic uses.
Sources
An AI photo editor is limiting its rights over the source image to the processing job
PhotoGPT updated its terms on September 24. Its tools include AI headshots, clothing and hairstyle changes, image enhancement, product photography and other generative editing. The terms say users retain ownership of uploaded images and generated outputs and grant PhotoGPT a licence solely to process the images to provide the service. The same terms prohibit non-consensual deepfakes depicting real people in sexual or compromising situations and material that infringes copyright or privacy rights.
Why visual creators should care
For photographers and agencies, 'you retain ownership' is much more useful when the accompanying platform licence is narrow enough to explain what the service may actually do with the file.
Practical takeaway
When reviewing an AI photo service, read the licence immediately after the ownership statement. Retaining copyright means less if the following clause gives the service broad rights to reuse, publish or train on the photograph.
What to watch next
Watch whether PhotoGPT's privacy documentation gives equally precise information about retention, third-party model processors and model-training treatment.
Sources
A stock library mixing photography and AI imagery is putting clearance back on the contributor
Anchorly's September 24 terms cover a free visual library containing both conventionally photographed and AI-generated or AI-assisted images. Contributors must represent that they created or hold the rights to submitted work, have obtained necessary model and property releases and do not infringe copyright, trademark, privacy or publicity rights. For AI-assisted material, contributors additionally warrant lawful access to the AI tool and that the work does not replicate identifiable copyrighted works, public figures or protected trademarks. Contributors grant Anchorly a worldwide, royalty-free, non-exclusive licence to distribute and sublicense accepted content under its own Ministry License.
Why visual creators should care
Mixed human/AI libraries increasingly need one contributor workflow capable of handling very different provenance chains. A camera photograph may require model and property releases; a generated image can add model-provider terms, source-material and synthetic-likeness questions on top.
Practical takeaway
For stock submissions, keep the clearance record with the asset. AI-assisted work can require more documentation than a conventional photograph, not less.
What to watch next
Watch whether smaller stock libraries begin asking contributors to disclose the generator and model version rather than relying only on broad warranties that the submitted AI work is lawful.
Sources
Another deepfake case begins with ordinary photographs of real people
Police in Henderson, Kentucky say a middle-school student is facing charges after allegedly using photographs of other students to create explicit AI-generated images by placing their faces onto different bodies and then showing the resulting material to others. Because the people involved are juveniles, authorities have released limited identifying information.
Why visual creators should care
The underlying visual-rights point extends beyond this case. A photograph can be lawfully accessible while the synthetic depiction subsequently created from the person's face is entirely unauthorised. Access to the source image and consent to the new representation are separate questions.
Practical takeaway
For professional portrait and likeness workflows, make synthetic transformation an explicit permission. Do not infer permission to generate a new depiction merely because a client, photographer or agency legitimately possesses the original portrait.
What to watch next
Watch how US state laws and school policies increasingly distinguish possession of an ordinary photograph from creation and distribution of synthetic intimate imagery.
Sources
