This video essay examines how generative AI reshapes authorship in contemporary media and asks what happens to the idea of the ‘death of the author’ when AI is used to create and distribute images and XR projects (Powers, 2016). After outlining current transformations and likely developments over the next five years, it advances a three-part argument about creation, interpretation and responsibility: AI either operates as a bounded tool or as a new author, but in neither case does authorship disappear, and human responsibility does not evaporate (Watson, 2025).

In Group D’s sharing on XR and AI co-creation, generative models are integrated with motion tracking and real-time engines so that environments can be generated and modified live. This accelerates pre-production and lowers entry barriers, yet tends to distribute creative control across model providers, production companies and performers, making it harder to determine whose contribution counts as authorship (Lee, 2024).

When face and voice replacement tools are folded into the same workflow, what appears to some as enhancement may be experienced by the performers being replaced as manipulation or loss of control (Ramos-Zaga, 2025). Within the next five years these systems are likely to be tightly coupled with user-behaviour analytics and biosensing, generating XR narratives that adapt to audience responses, while legal and regulatory pressure pushes towards standards for watermarking synthetic material, logging prompts and disclosing AI assistance (Ramos-Zaga, 2025).

In academic publishing, current policies generally permit language models to assist with drafting or polishing but forbid listing them as authors, and require human writers to disclose AI use and bear ultimate responsibility for arguments and citations (Lund & Naheem, 2023). Professional guidelines consolidate this tool-based positioning, calling for transparent procedures, audit trails and structured authorship statements rather than recognition of AI as an autonomous creative subject (Teixeira da Silva & Tsigaris, 2023).

This case shows how one influential domain is reasserting human authorship and responsibility at the end of AI-assisted workflows; over the coming five years, AI-assisted drafting is likely to become routine, accompanied by stricter checks on citations and screening for fabricated material (Watson, 2025).

At the level of creation, these practices show why AI has never brought about the ‘death of the author’ (Murray, 2024). When creators have clear goals, AI functions as a bounded tool: prompts guide generation, and humans select and arrange outputs, so authorship is reinforced. In the auntiverse case, the artist Niceaunties sets a loose framework, lets the model run, and then curates from the outputs elements that fit the project; authorship is redistributed across prompting, selection and curation but still present.

At the other end of the spectrum, some workers almost completely accept whatever the model outputs, with little prior conceptualisation and minimal revision; here there is no meaningful author who can die, because no one truly takes responsibility for the work (Teixeira da Silva & Tsigaris, 2023).

At the level of interpretation, the logic of AI creation also sits uneasily with the claim that the author should disappear and the reader occupy centre stage (Powers, 2016). The slogan of the ‘death of the author’ was meant to challenge authorial authority and defend reader freedom (Dickie & Wilson, 1995), yet when creators treat AI as a bounded tool, outputs remain constrained by human prompts, training choices and selection; when AI is allowed to generate freely, it becomes a new author with opaque authorship (Gunkel, 2012).

In both situations, interpretation is never entirely independent of the author (Carroll, 1997). Everyday internet use makes this clearer: users appear to wander among limitless resources, but what they see is shaped by algorithmic ranking, content moderation and interface design, so every feed and search results page carries an author-function (Donig, 2021).

Some posthuman accounts of AI claim that humans and non-humans share agency and that responsibility is distributed across the network (Watson, 2025). Such system-level descriptions have analytical value, but once concrete harms occur this way of speaking becomes ethically hazardous (Teixeira da Silva, 2023). Research in information systems and law still reserves the status of ‘person’ and responsible agent for humans, insisting that only human individuals and institutions can hold rights, bear obligations and be sanctioned (Moulaison-Sandy, 2023).

A responsibility-centred anthropocentric position therefore remains necessary. Creative work today is produced by assemblages of humans and machines, but design decisions, training choices and deployment strategies must still be traced back to identifiable artists, studios, publishers and platforms, so that responsibility does not quietly vanish into the machinery (Lee, 2024).