UK creators have pushed training-data disclosure past the 10,000-signature mark
A UK Parliament petition calling for a statutory duty requiring AI developers to disclose the copyrighted works used to train generative AI has passed 10,000 signatures. The petition asks for disclosure detailed enough for creators to identify whether their work was used, enforce their rights, negotiate licences and seek payment. It remains open until October 29.
Why visual creators should care
Training transparency is one of the practical problems photographers and illustrators face when trying to exercise rights over work that may have entered a model. A creator cannot meaningfully negotiate, object or pursue payment if they cannot establish whether their work was used in the first place.
Practical takeaway
Keep organised records of published portfolios, image identifiers, copyright registrations where applicable and licensing history. Training-data disclosure is only useful if a creator can match disclosed material back to work they can prove they control.
What to watch next
Watch the UK Government's formal response and whether the petition gains enough support to increase pressure for parliamentary debate or feed into the broader UK copyright-and-AI policy process.
Sources
A Delhi court is asking where protection from AI likeness abuse should actually stop
The Delhi High Court has granted limited interim relief in Janhvi Kapoor's personality-rights case involving allegedly pornographic AI-generated material, impersonation accounts and other online uses of her identity. Importantly, the court rejected the idea of simply removing everything covered by the much broader request. Justice Anup Jairam Bhambhani questioned blanket action against fan pages and also raised the relationship between personality rights and the separate copyright held by a photographer, artist or other creator of an image.
Why visual creators should care
Likeness protection can collide with other visual rights. Someone may have a strong claim against a sexual deepfake made without consent while a photographer may separately own a perfectly lawful portrait of the same person. Treating every visual use of a person's identity as the same thing risks erasing that distinction.
Practical takeaway
For photographers and agencies, keep model permission and image copyright as separate records. A subject's personality or likeness rights do not automatically answer who owns a photograph, while owning the photograph does not automatically authorise every synthetic transformation of the person depicted.
What to watch next
The case is due back before the Delhi High Court on August 17. Watch whether the court develops a clearer distinction between obscene synthetic misuse, commercial impersonation, ordinary photography, fan expression and lawful copyrighted works.
Sources
WIPO is telling AI users to read the contract, not just the copyright headline
WIPO's 2026 Learning Machines guide includes detailed practical guidance on intellectual property in AI terms of service. It highlights three recurring issues across text, image, audio and video generators: ownership of inputs and outputs, liability when generated material infringes copyright and platform moderation. WIPO notes that providers may assign output rights to users while retaining broad reuse rights themselves, and that infringement liability is often pushed back onto the user.
Why visual creators should care
A creator can have permission to use an AI tool without necessarily getting exclusivity, confidentiality or meaningful protection if something goes wrong. For client work, the tool contract can therefore matter as much as the copyright status of the final image.
Practical takeaway
Before using a generator commercially, check who can reuse your input and output, whether uploads can be used for training, who carries infringement liability and what disclosure the client expects. WIPO also recommends keeping records showing human creative decisions where authorship matters.
What to watch next
Watch whether professional procurement increasingly turns WIPO's basic questions into standard AI clauses covering no-training commitments, output licences, indemnity, disclosure and retention of project evidence.
Sources
Creators are beginning to write their own AI production rules
Fresh reporting from Business Insider shows audience resistance to AI increasingly affecting how creators define their production process. After criticism over his use of AI as a research aid, science creator Hank Green published a policy saying scripts would not be written, edited or outlined by an LLM, the thesis of a video must originate with a human, and no image or music in the video would be AI-generated. The same report describes backlash against creators promoting AI companies or using generative systems heavily in finished work.
Why visual creators should care
AI disclosure is becoming a client and audience expectation rather than simply a legal requirement. For photographers, filmmakers and agencies, a clear production policy can be more useful than vague statements such as 'AI-assisted', because it tells people where the boundary actually sits.
Practical takeaway
Write down what your own studio means by AI-assisted work. You might allow research, masking or cleanup while prohibiting generated final imagery, synthetic people or generated client assets. The important thing is that the rule is clear enough to explain and record.
What to watch next
Watch whether creators, agencies and brands begin publishing more explicit AI production policies and whether client briefs move from generic 'no AI' language towards distinctions between generative creation and ordinary AI-assisted editing.
Sources
One photography competition is solving the AI argument by giving generated work its own category
The 2026 Global Photography Awards explicitly lists AI-Generated Photography as a special category, separate from categories including editorial, fashion, commercial, film and fine-art photography. Its final-extension entry period begins August 14. The competition describes the special section as a place for experimental practices that push the boundaries of traditional photography, including artificial intelligence.
Why visual creators should care
Photography competitions have taken very different approaches to generated imagery, from outright bans to loosely worded restrictions. A separate category is another model: generated work is accepted, but entrants do not need to pretend that it belongs to exactly the same production category as camera-made work.
Practical takeaway
When entering competitions, do not assume the word 'photography' tells you whether AI is permitted. Read the category rules and submit generated work in the category intended for it. Keep the rules that were in force on the submission date.
What to watch next
Watch whether more competitions move towards clearly separated camera, AI-assisted and AI-generated categories rather than trying to force every visual workflow into a single definition.
Sources
The National Portrait Gallery is showing what an opt-in synthetic portrait workflow can look like
The National Portrait Gallery's ongoing A National Portrait project, created by Es Devlin with Google Arts & Culture Lab, lets adults in the UK photograph their own face and have it transformed into an animated portrait based on Devlin's charcoal and chalk visual language. Participants can download the result and separately choose whether it joins the evolving gallery installation. The Gallery directs participants to terms explaining how their data will be processed before they submit.
Why visual creators should care
Much of the current likeness debate concerns images generated from people who never agreed to participate. This project provides a very different production model: the artist is part of the collaboration, the person supplies their own face knowingly and public display is an additional choice.
Practical takeaway
For commercial synthetic portrait projects, break consent into stages. Permission to upload a likeness, permission to generate from it and permission to publish the resulting image do not have to be one blanket approval.
What to watch next
Watch whether museums, agencies and commercial campaigns adopt similarly explicit opt-in structures for AI portraits and whether participant-facing terms become easier to understand as synthetic likeness workflows spread.
Sources
