Information

ACC-R is an experimental AI model agency and speculative design project that examines what happens when artificial intelligence enters the representational systems traditionally reserved for human identity. This project creates and manages a growing roster of AI-generated models. These are not real people. They are synthetic constructs produced through generative image systems, entered into a formal registry, each model exists as a replica: a stand-in for something that was never there to begin with.

Fashion and commercial imagery have always traded in constructed identity. Models become surfaces for projection, vessels for brand narratives, lifestyle aspirations, and cultural codes. The industry has spent decades refining systems for cataloguing, representing, and circulating human appearances. Agencies maintain portfolios. Clients book talent. Images enter archives.

ACC-R builds synthetic models using current generative AI tools. Each model is developed as a coherent visual identity, consistent appearance and defined characteristics, so far as the current technology allows. They function as creative assets available for projects that require human-like imagery without human subjects. The agency exists because the tools now exist. Image generation has reached a threshold where synthetic models can serve real production needs. ACC-R tests that capability by operating as a functioning entity rather than a theoretical proposition.

There are open questions about AI-generated human likenesses. Questions about ethics, labour, authenticity, creative value. These questions will not be resolved through observation. They require contact. They require working with the material, encountering the friction, finding the edges. This project engages by making. By building models. By putting them into circulation. By seeing what works, what fails, what feels right, what feels wrong. Is this ethical? That is part of what the experiment examines. Not from the outside, but from within the work itself. The position is honest: this territory is blurred, the boundaries are unsettled, and the only way to understand it is to operate inside it. Is this a legitimate form of creative expression? Is it a new mode of working? Is it something else entirely? This project does not presuppose the answers. It generates the conditions where answers might emerge.

This is not a position of neutrality. It is a position of active engagement with difficulty. The project proceeds with awareness that the ground is unstable, and proceeds anyway, because instability is where understanding develops. Some will view this work as problematic. Some will view it as inevitable. Some will see creative possibility. Some will see ethical breach. This project exists in the space where these perspectives collide.

There is a possibility that synthetic model generation represents something genuinely new, not a replacement for existing practice but an expansion of creative capacity.

Imagery without the constraints of physical production. Models without scheduling conflicts, ageing, or geographical limits. Visual identities that can be purpose-built for specific projects, specific aesthetics, specific needs. This could be a tool. A collaborator. A new category of creative asset that sits alongside photography, illustration, and CGI as a distinct production method with its own characteristics and applications.

Accession. The act of formally entering something into a registry. An acknowledgment that what is being created has enough substance to warrant documentation and organisation. Replica. A representation. Not an original, not pretending to be. A functional stand-in that serves a purpose while being clear about its nature.

This is a free and open catalogue, all images are copyright free and commercially available, there will never be a charge on the content produced via this project. For further information or collaborations contact via socials.

Project Roadmap:

  • Initial test model group M-A released.
  • Produce and diversify representative model grouping.
  • Create downloadable content packages for model hiring.
  • Open prompt library for generative editorials.
  • Open papers on findings and tools.
  • More coming soon...