Salford researcher contributes to new guidance on use of AI within filmmaking

Dr Daniella Gáti, a Lecturer in Interactive Media, contributed to the document ‘Guidelines for the Sexism-Free and Gender-Sensitive Use of AI in the Film and Media Industries’, that was launched by feminist think tank Power To Transform! at the 2026 Berlinale (Berlin International Film Festival).

The guidelines bring together expert knowledge and perspectives from a range of academics and also provide a list of key recommendations for filmmakers in addition to a practical toolkit and hands-on checklists.

Dr Gáti’s contribution to the guidelines, titled ‘Queer Identity, Algorithmic Knowledge, and the Conservative Nature of AI’ sets out the following points for filmmakers:

•    That queer characters in film will appear in stereotypical forms if content is generated by AI, due to how they have historically been either absent from film history or reduced to the function of villainous characters or sidekicks. Dr Gáti states that the only way to combat this would be through human intervention, calling for more complex representations of queer characters on screen to provide more data for AI to draw from. They also note that the games industry has a bigger role to play here, because it has a historically ‘misogynistic culture’ and has often marginalised women and non-binary people in its content.

•    Unless you explicitly ask for non-normative characters such as queer characters in prompts to generative AI systems, you will not get them. Dr Gáti explains this is a phenomenon called ‘quiet normativity’ and urges filmmakers to avoid using generative AI when building narratives or designing characters, or at least to reflect on AI outputs before adopting them into plot and character design.

•    That generative AI is a deeply misleading term as Dr Gáti states that AI is ‘fundamentally conservative.’ This is because AI does not create content that is actually new as when humans invent things, but rather relies solely on identifying patterns in existing data to recombine them in ways that reproduces what already exists. This is also the reason that AI cannot imagine futures.

•    Dr Gáti introduces the term ‘algorithmic erasure’. This occurs when AI produces outputs based on massive datasets, each broken down into distinct categories, in which not all of them are identifiable. As this form of statistical categorisation struggles to account for gender fluidity, which defies categories, it means that AI systems are incapable of representing fluid identities and then reinforce strictly separate identity categories—of gender, sexuality, but also of other identities such as race, class, or political affiliation.
 

Dr Gáti said: “With generative AI and filmmaking, I think that the industry should hesitate, learn more, and reflect on how AI works and what its consequences are before rushing into using it for its practices.
 

“In film, television, and games, stronger storytelling comes from deeper creativity, not speed alone. I understand the pressure creative practitioners face to deliver fast results — the fear that if you do not use AI, someone else will. Still, I urge people to resist where possible.”

The guidelines document includes other contributions from academics from the University of Cambridge, The International Central Institute for Youth and Educational Television, Munich, New York’s Hunter College and Ada Developers Academy in Seattle.

The full guidelines document is available at the Power to Transform! website.