[
  {
    "title": "Time to ACCCT: Providing Creative Industries and AI Developers with a Copyright Framework of Access, Control, Consent, Compensation and Transparency",
    "authors": [
      "James Bennett",
      "John Collomosse",
      "Rebecca Gregory-Clarke",
      "Jack Jones",
      "Lynn Love",
      "Mark Lycett",
      "Will Saunders"
    ],
    "venue": "CoSTAR/DECaDE Technical Report",
    "year": 2025,
    "summary": "This report introduces the ACCCT (Access, Control, Consent, Compensation, Transparency) framework as a technical foundation for trustworthy data and content flows in generative AI. It highlights the need for auditable data supply chains and proposes mechanisms for embedding provenance, attribution, and licensing directly into digital assets through standards such as C2PA, watermarking, and tokenised licensing. Time to ACCCT demonstrates how technical interventions can operationalise creator rights, enable transparent data use, and provide a secure infrastructure for responsible AI development. The report was developed through several workshops with UK creative industries stakeholders and AI developers in the wake of the UK government consultation on Copyright and AI.  It concludes with recommendations that support a provenance backed content licensing framework supported by open standards as a practical solution.",
    "image": "assets/img/accct_report.png",
    "links": {
      "pdf": "pubs/Bennett-ACCCT-2025.pdf",
      "bib": "pubs/bib/Bennett-ACCCT-2025.bib"
    }
  },
  {
    "title": "Multitwine: Multi-Object Compositing with Text and Layout Control",
    "authors": [
      "Gemma Canet Tarres",
      "Zhe Lin",
      "Zhifei Zhang",
      "He Zhang",
      "Andrew Gilbert",
      "John Collomosse",
      "Soo Ye Kim"
    ],
    "venue": "CVPR (Highlight)",
    "year": 2025,
    "summary": "We introduce the first generative model capable of simultaneous multi-object compositing, guided by both text and layout. Our model allows for the addition of multiple objects within a scene, capturing a range of interactions from simple positional relations (e.g., next to, in front of) to complex actions requiring reposing (e.g., hugging, playing guitar). When an interaction implies additional props, like `taking a selfie', our model autonomously generates these supporting objects. By jointly training for compositing and subject-driven generation, also known as customization, we achieve a more balanced integration of textual and visual inputs for text-driven object compositing. As a result, we obtain a versatile model with state-of-the-art performance in both tasks. We further present a data generation pipeline leveraging visual and language models to effortlessly synthesize multimodal, aligned training data.",
    "image": "assets/img/multitwine.png",
    "links": {
      "pdf": "pubs/CanetTarres-CVPR-2025.pdf",
      "bib": "pubs/bib/CanetTarres-CVPR-2025.bib"
    }
  },
  {
    "title": "To Authenticity, and Beyond! Building Safe and Fair Generative AI upon the Three Pillars of Provenance",
    "authors": [
      "John Collomosse",
      "Andy Parsons"
    ],
    "venue": "IEEE Computer Graphics and Applications (CG&A)",
    "year": 2024,
    "summary": "Provenance facts, such as who made an image and how, can provide valuable context for users to make trust decisions about visual content.  Against a backdrop of inexorable progress in Generative AI for Computer Graphics, over two billion people will vote in public elections this year.  Emerging standards and provenance enhancing tools promise to play an important role in fighting fake news and the spread of misinformation.  In this paper we contrast three provenance enhancing technologies:  metadata, fingerprinting and watermarking, and discuss how we can build upon the complementary strengths of these three pillars to provide robust trust signals to support stories told by real and generative images.  Beyond authenticity, we describe how provenance can also underpin new models for value creation in the age of Generative AI.  In doing so we address other risks arising with generative AI such as ensuring training consent, and the proper  attribution of credit to creatives who contribute their work to train generative models.  We show that provenance may be combined with distributed ledger technology (DLT) to develop novel solutions for recognizing and rewarding creative endeavour in the age of generative AI.",
    "image": "assets/img/ieeeprov.png",
    "links": {
      "pdf": "pubs/Collomosse-IEEECGA-2024.pdf",
      "bib": "pubs/bib/Collomosse-IEEECGA-2024.bib"
    }
  }
]