In this regular feature, we round up this month’s key legal developments in AI that brand owners should know about, and explain what they could mean for your business.

October updates

UK

The ICO raises the bar on generative AI transparency

The ICO has published a report, Building trust and transparency into generative AI development (October 2026), setting out what it expects of AI developers. It finds continuing, and in some cases worsening, shortfalls in how developers explain their use of personal data. The report draws on a three-year supervision programme covering the 11 largest foundation model developers active in the UK, including OpenAI, Google, Meta, Microsoft and Stability AI. 

The report applies directly to any organisation that deploys, fine-tunes or buys generative AI tools, which will soon include almost every business. These expectations pass down the supply chain, so it is worth checking how open your suppliers are about their use of generative AI.

AI use by lawyers

The Law Society has published a practice note, Using artificial intelligence in legal practice, with guidance for solicitors on adopting AI tools responsibly. It highlights the many tasks AI tools can help with, but rightly stresses the importance of choosing the right tool and the risks of relying too heavily on it. For in-house lawyers, the note is a useful benchmark, both for weighing the benefits and risks of AI tools and for asking external advisers how they use AI on your matters and what safeguards they have in place.

Digital Twins Bill 

On 9 September 2026, Dame Chi Onwurah MP, Chair of the Science, Innovation and Technology Committee, introduced the Personal Data (Digital Twins) Bill. The Bill would regulate "digital twinning": software or algorithms that use or hold personal data to model a person's preferences or behaviours, including content creation algorithms, deepfakes and chatbots. It would not ban digital twins but would require explicit consent before one is created. Children would be expressly protected as parental consent could not substitute for their own, and unauthorised replication of a person's likeness would be prohibited. The Bill’s second reading is scheduled for 13 November 2026.

Oxford Consultation on Copyright and AI

The University of Oxford's Institute for Ethics in AI has published The Oxford Consultation on Copyright and AI (October 2026), a cross-sector report that calls on the UK Government to act on the unauthorised use of copyright works in AI training. 

Its recommendations sit under three pillars: ownership and consent, transparency, and fair licensing. It rejects text and data mining exceptions and opt-out regimes. It also opposes a government-administered levy. 

It says the High Court's ruling in Getty v Stability AI shows a gap in secondary infringement law. To close it, the report proposes removing the knowledge requirement for secondary infringement. Its other proposals include mandatory disclosure of training data, a designated regulator, and statutory penalties alongside court proceedings.

Upcoming…

Copyright and AI training go before the Court of Appeal

The UK's leading copyright case on generative AI, Getty Images v Stability AI, has reached the Court of Appeal, with the hearing due on 11 November 2026. At first instance, the High Court found that Stability AI was not liable for secondary copyright infringement, holding that the model weights behind its Stable Diffusion image generator were not an "infringing copy" of the images used to train it. Getty did win a limited trade mark claim where generated images reproduced its watermarks.

The appeal matters well beyond the parties involved. In granting permission to appeal, the judge described the legal point as "both novel and important", with possible implications for intangible articles such as software generally. If Getty succeeds, bringing foreign-trained AI models into the UK, for example by downloading them, could amount to UK copyright infringement. 

The appeal carries extra weight because of the Government's position on copyright and AI. In its March 2026 report, the Government dropped the idea of a broad commercial text and data mining exception. Instead, it chose to gather evidence, set up industry working groups and pilot a "Creative Content Exchange" for licensing. That leaves the Getty case as the main test of what businesses can lawfully do in the UK.

Have your say on agentic AI

The next frontier is likely to be "agentic" AI: systems that can act on their own rather than simply answer questions. The ICO has issued a call for evidence on agentic AI, which runs until 20 November 2026.

EU

First harmonised standard supporting the AI Act 

The European Commission has adopted the first harmonised standard supporting the AI Act. The standard gives providers of high-risk AI systems a clear route to meeting the quality management system requirements under Article 17 of the AI Act. Compliance with a harmonised standard published in the Official Journal gives rise to a presumption of conformity with the corresponding requirements.

Upcoming…

EU AI Act deadlines

Two EU AI Act deadlines fall on 2 December 2026. Article 50(2) requires providers of AI systems that generate synthetic audio, image, video or text to mark outputs in a machine-readable format detectable as artificially generated or manipulated. 

The Digital Omnibus on AI (Regulation (EU) 2026/1744) gives systems placed on the market before 2 August 2026 until 2 December 2026 to comply. Systems launched on or after 2 August had to comply from launch. The grace period is narrow. It does not postpone deployers' duties, such as disclosing deepfakes, which have applied since 2 August 2026. Also, AI systems built to create child sexual abuse material or non-consensual intimate imagery become prohibited. 

Copyright consultation

The European Commission's targeted consultation on copyright and AI closes on 3 November 2026. 

It covers: licensing, enforcement and performer imitations in generative AI; piracy of live and other time-sensitive content; introducing material reciprocity for the single equitable remuneration right, where non-EU countries do not offer equivalent rights to EU performers and producers; and uneven national implementation of the optional scientific research exception under the Copyright Directive (2001/29/EC).

Like Company v Google

The Advocate General’s opinion is due on 3 November 2026. This is a claim brought by Hungarian news publisher Like Company against Google in relation to outputs generated by Google’s Gemini chatbot.

US

Cases and enforcement 

October was another busy month in the US courts. 

In Thomson Reuters v Ross Intelligence, the Third Circuit became the first US appellate court to reject a fair use defence to AI training, after Ross copied Westlaw content to build a competing legal research tool. The ruling is narrow: Ross's tool was not generative AI, and the court left open how its reasoning would apply to generative models.

For brand owners, a notable new claim is the suit brought against Anthropic on 23 October by Authentic Brands Group, owner of the Elvis Presley catalogue and image rights, which is seeking around $41 million in respect of 274 songs. Separately, Dow Jones and the New York Post have added trade mark claims for false designation of origin and dilution to their copyright case against Perplexity.

Commentary

New research by NYU Law's Barton Beebe and Jeanne C. Fromer, co-written with David Stein of Vanderbilt Law School, is the first large-scale computational study of how the US Patent and Trademark Office (PTO) handles image trademarks. The article is titled "Trademark Law's View of the Image: A Computational and Empirical Analysis." 

It uses AI to examine 2.9 million image-based trademark applications filed with the PTO between 1985 and 2023. The authors' findings suggest that decades of logo registrations rest on a surprisingly shaky foundation. 

Other territories

WIPO

Over 970 judges registered for the 2026 IP Judges Forum, held on 14–15 October 2026. AI featured in three of the nine sessions, covering AI-assisted outputs, the use of works to train models, and AI in court proceedings, with the training discussion drawing on GEMA v Suno, Bartz v Anthropic and ANI v OpenAI. A further session addressed comparative and referential use of trade marks. Why it matters: rather than seeking a single global answer, judges are comparing their approaches to reasoning, so brand owners should expect outcomes to continue to vary between jurisdictions.

WIPO's first Global Forum on IP and AI runs on 9–10 November 2026 and covers trade marks, copyright and designs. 

If you would like to discuss how these developments affect your business, please get in touch with your usual Stobbs contact.