AI in production · The record
The published record, one line each.
Short, dated and linked, with anything still moving marked as such.
Each line was checked on 8 November 2026 against the organisation's own page or a named report. Nothing here is legal advice.
Industry bodies and unions
| Who | What they've published | Date |
|---|---|---|
| BFI, with the CoSTAR Foresight Lab | AI in the Screen Sector: Perspectives and Paths Forward: nine recommendations, followed by an AI Casebook of UK case studies and a quarterly AI Digest. BFI report hub. | June 2025 |
| Pact | Five principles for members on generative AI: respect copyright, value human creativity, responsibility and transparency, diversity, and data privacy. Pact. | 13 March 2024 |
| Pact and Equity | After Equity's ballot, Pact sent improved AI wording on 19 January 2026 and both sides agreed to keep negotiating. Talks continued in June 2026; no published agreement at our last check. BroadwayWorld, 21 January 2026. | January 2026 |
| Equity | "Stop AI Stealing the Show": consent, transparency and payment for the use of performers' voice, face and body. In an indicative ballot in December 2025, 99.6% of members who voted said they'd refuse to be digitally scanned on set. Equity. | December 2025 |
| Bectu | Welcomed the government dropping its AI copyright opt-out plan, and wants consent, credit and fair pay in whatever comes next. Bectu. | 18 March 2026 |
| BBC | Published AI principles: act in the public's interest, prioritise talent and creativity, and be open and transparent about where AI is used. BBC. | Updated 7 February 2024 |
| AIMICI and ScreenSkills | ScreenSkills-accredited Responsible AI training: level 1 is about an hour, level 2 four to five hours. Pact is the launch partner for production companies. Pact. | 15 June 2026 |
Government and law
| Who | What it says | Date |
|---|---|---|
| UK government | Report on Copyright and Artificial Intelligence: a broad copyright exception with an opt-out "is no longer the government's preferred way forward". It proposes more evidence-gathering, work with industry on labelling, and exploring a possible new digital replica or personality right. GOV.UK. | 18 March 2026 |
| European Union | AI Act Article 50: transparency duties, including disclosure of deep fakes and machine-readable marking of synthetic content. They apply from 2 August 2026; the AI Omnibus (in force 27 July 2026) didn't defer them, and systems already on the market have until 2 December 2026 for marking. Article 50 text; White & Case on the Omnibus. | August 2026 |
| Courts: Getty Images v Stability AI | The High Court rejected Getty's secondary copyright infringement claim, finding the Stable Diffusion model was not an "infringing copy". Getty was given permission to appeal on that point in December 2025. We found no Court of Appeal judgment at our last check. CMS case tracker; Burges Salmon. | Judgment 4 November 2025; appeal pending |
What this adds up to
Nobody credible is against AI, and nobody is unconditional. The shared vocabulary is human creativity, consent, credit, transparency, copyright and data, and it's the vocabulary of our position and course.
If something here has moved, tell us at [email protected].
Last reviewed 9 November 2026