British Authors Debate Ethical AI: What the Society of Authors AGM Means for Writers
British Authors Debate Ethical AI: What the Society of Authors AGM Means for Writers
Friday 20 November 2026 | AI & The Author
The Society of Authors discussed AI at its 19 November AGM. What is known, what remains unconfirmed, and what writers should do now.

A defining conversation for British writers
On Thursday 19 November 2026, the Society of Authors held its annual general meeting online. Its published agenda explicitly included discussion of artificial intelligence in publishing and the organisation’s role in responding to it. That is the verified development at the centre of today’s edition. As of this morning, I have not found an official account of the discussion or any decisions reached. It would be misleading to present the agenda as a vote or claim that the Society adopted a new policy. Even so, the meeting matters because authors are confronting difficult questions about consent, creativity, contracts and the limits of AI assistance.
Why this meeting matters now
The AGM followed a turbulent few weeks for the Society’s approach to industry standards. On 14 October the organisation announced its involvement in a Charter of Ethical AI Use in the Creative Industries. On 21 October it withdrew from that charter, saying it needed time to listen to members’ concerns. It specifically identified the November AGM as an opportunity to discuss next steps. That sequence explains why a scheduled conversation about AI was more than a routine agenda item. It placed disagreement about acceptable uses of technology at the heart of a professional body representing writers.
The charter question
An ethical charter can sound reassuring, but its value depends on what it actually requires. Authors may agree that AI should be used responsibly while disagreeing about what responsibility means in practice. Is machine-assisted proofreading acceptable? What about translating a manuscript, generating marketing copy, producing an audiobook voice or drafting narrative passages? A single label cannot resolve these distinctions. The practical questions are who has control, whose work is used, what permissions have been granted and whether readers are given an accurate account of the creative process.
What is confirmed and what is not
The Society’s AGM notice establishes the date, time and intended AI discussion. Its October update establishes that it withdrew from the charter pending member feedback. Neither source establishes what members said on 19 November, whether a revised charter was proposed or what the organisation will do next. Those outcomes require a subsequent official statement or minutes. This distinction is particularly important in a daily news article: the absence of confirmed results is not a licence to invent a dramatic resolution.
The UK copyright position
The UK government’s March 2026 report on copyright and artificial intelligence remains essential background. Ministers moved away from their earlier preference for a broad text-and-data-mining exception with a rightsholder opt-out. The report did not itself enact a replacement licensing regime or settle every legal question about AI training. The government described further work on creator control, transparency and other issues. For authors, that means practical rights management still matters while policy develops. A political statement, an industry charter and binding law are three different things.
Why consent comes first
The strongest argument from many writers is not that useful software should be prohibited. It is that original work should not become a free raw material simply because it is digitally accessible. Publishing a novel, essay or poem is not necessarily an invitation to train a commercial system on it. Authors need workable ways to understand where their work goes and to negotiate its use. Developers, meanwhile, argue that broad access to information can improve tools and competition. Sustainable rules must acknowledge both innovation and the economic value of creative labour.
The risk of broad declarations
A pledge to avoid AI can become misleading if it fails to distinguish ordinary tools from substantial machine-generated expression. Spellcheckers, accessibility features, transcription and research discovery already form part of many writers’ workflows. The more useful boundary is between assistance that supports an author’s decisions and a system that supplies the creative substance. Readers should not have to guess which is which. Nor should authors be pressured into claiming that every stage of their work was entirely free of automation when that is neither necessary nor accurate.
Contracts need more precision
For authors negotiating with publishers, agents or production partners, the immediate task is to examine contractual language. Agreements may need to address training licences, manuscript uploads, AI narration, translation, derivative rights and the use of an author’s name or voice. A clause that simply mentions artificial intelligence may be too vague to protect either party. Writers should ask which uses are permitted, who approves them, whether compensation applies and how consent can be withdrawn where legally possible. Professional legal advice remains important for individual agreements.
Independent authors face different pressures
Self-published writers often control their own production, but that freedom comes with responsibility for choosing tools and protecting files. They may rely on AI-assisted administration to compete with larger businesses while still insisting that their storytelling remains human-led. They also have fewer resources for monitoring unauthorised copying or pursuing disputes. Industry standards will be credible only if they work for independent novelists and small publishers, not merely organisations able to employ specialist legal and technical teams.
Human authorship and reader trust
The Society of Authors launched its Human Authored labelling initiative earlier in 2026. Its published guidance is intended to help readers identify books that meet the scheme’s definition of human authorship. Such labels can support informed choices, but they are not substitutes for copyright law, fair contracts or a careful assessment of individual works. A label is most valuable when its criteria are transparent and consistent. The wider challenge is preserving confidence in writing without encouraging unproven accusations against authors who happen to use a particular style.
A practical checklist for this week
Keep dated versions of your manuscripts and a record of substantial editorial changes. Review the terms of writing and publishing platforms before uploading complete drafts. Identify any clauses that might allow training or synthetic reproduction of your work. Decide in advance how you will describe permitted AI assistance to readers. If you are a Society member, consult official AGM communications rather than second-hand accounts. These measures will not settle the national debate, but they put authors in a stronger position to respond.
My editorial view
I believe authors should retain control over the central creative decisions in their work and should be able to choose useful technology without surrendering their intellectual property. Ethical AI policy will only succeed if it is specific enough to protect real people and flexible enough to distinguish assistance from replacement. The Society’s willingness to hear objections matters. What happens after that discussion will matter more. Clear definitions, meaningful consent, fair compensation and honest disclosure are more valuable than slogans from either side.
What to watch next
The next verified milestone will be the Society’s own account of the AGM discussion and any proposed follow-up on the withdrawn charter. Authors should also watch the government’s continuing work on transparency and creator control, and the practical development of licensing markets. For now the headline is deliberately measured: British authors gathered to debate ethical AI, but no outcome should be attributed to them without evidence. That is a more useful starting point for responsible coverage than an invented victory or defeat.
About Rob
Rob Frankson writes AI & The Author at 121minutes.com alongside his original science-fiction work. The column examines authorship, emerging technology and the changing business of writing.
Subscribe and editorial transparency
Explore more articles and subscribe through 121minutes.com. This article was prepared with AI assistance for research discovery and drafting, and requires the author’s editorial review before publication. AI assistance does not replace responsibility for factual accuracy, original creative judgement or rights clearance.
Sources and useful shortcuts
Society of Authors AGM agenda (19 November)
Society of Authors October charter update
.png)









Comments