AI is everywhere as we go into 2026. 2025 saw AI rise up the global agenda to take a seat at the top table of geopolitical issues. 2026 will see a rush to AI integration that adds genuine value to business around the world. Whether this will bring ‘revival, bust or backlash’ in 2026 (in the words of The Economist) remains to be seen.[1]
Bridging the AI investment gap between cost and returns
The gap between annual global AI investment (estimated to reach $1trn by 2030) and returns (currently $50bn) means there will be a huge push towards successful AI adoption at scale in 2026. This is necessary in order to increase AI project success rates and get the economics working. There’s a way to go here: a survey by MIT in mid-2025 found that only 5% of corporate AI projects made a return.[2]
AI and the technical landscape
What’s driving the frenetic pace of development is scaling laws – the AI equivalent of Moore’s law where AI performance doubles every 6 months (instead of 24 months for processor performance under Moore’s law).
We’re all used to the Cloud Tech stack and the ‘as a service’ componentry of software (applications), platform (operating system, middleware) and infrastructure (servers, networking).
Essentially, AI and machine learning enrich each layer of the stack and add several new ones:
- agents and AI models (as app building blocks) sitting in the top (app server) layer;
- the data layer (underneath the apps layer) and including data sources, APIs, databases, analytics, business intelligence and governance; and
- AI accelerators, training and inference, accessed through Cloud infrastructure.[3]
AI and the business landscape
The business landscape at the moment is a tale of two cities. At the macro level, the Bank of England in an October 2025 blog noted that 44% of the market capitalisation of the entire S&P 500 index was accounted for by AI stocks.[4] It is these companies, together with AWS and leading foundation model providers OpenAI and Anthropic that have generated most headlines in 2025.
At the micro level, we’re at the ‘let 1,000 flowers bloom’ stage of the cycle, with a host of startup and enterprise businesses at every AI layer developing and commercialising applications.
AI and the legal services market
The legal services market is a Petri dish of AI-enabled development in a business sector undergoing wide-ranging structural change. Machine learning and AI have for a while helped take cost out of repeatable, standardisable tasks in larger legal projects, but what we’re now beginning to see is AI agents as team members starting to handle more and more aspects of the project under human guidance.
A search on Perplexity in November 2025 returned two dozen or so AI providers to the legal services market, focusing on specific vertical functionality (litigation e-discovery, M&A due diligence, property reports and compliance) and more general horizontal applications (research, document drafting, contract review and checklists) across the business. And there are doubtless as many again not on the Perplexity list.
As with the Internet boom in the late 1990s and the microcomputer boom of the early 1980s, there will at some point (whether in 2026 or later) be a consolidation phase which will radically change the landscape.
What does AI mean for lawyers in 2026: (1) contracting for AI
As AI in business moves from proof of concept to pilot to full roll-out, contracting for AI will get more complex in 2026. Against a backdrop of more pressure to see positive results there will be many more AI project agreements coming across Tech lawyers’ desks next year.
Papering proof of concept and pilot projects has been fairly straightforward. Moving to full roll-out, where the customer is actively relying on its AI for customer facing solutions, will be more challenging. AI service terms were materially pro-provider in 2023 and 2024 (long on customer obligations, short on customer rights). In 2025, as competitive pressures intensified, the market for AI started to behave more recognisably with contract terms becoming less one-sided. Increasingly, coming to market with ‘balanced’ T&Cs (sometimes benchmarked against peers by a third party) is a positive point of differentiation for AI vendors. Things to watch out for in AI contracts in 2026 from the customer standpoint include:
- Dependence and switching: your AI service is likely to be dependent on inputs from a wide variety of providers you have no direct contract with – what happens if a key provider to your service provider is removed from the supply chain? or bumps up its charges? Or your provider changes the models it uses for your service and this has a negative impact on performance?
- Accuracy: AI up to now just hasn’t been that accurate. What is an acceptable hallucination (accuracy) rate and how do you manage this?
- Where does your data go? Does it end up in a big Tech database somewhere without you knowing about it?
- Copyright: Does the provider expect an indemnity on your use of the service? Will it give you any comfort on the model infringing third party rights?
- Big Tech terms: as with cloud contracts, the scope for negotiation with big Tech providers will be small.
- Liability regimes: liability caps in cloud contracts are settling down at 1x to 2x annual charges, with higher caps (or uncapped) for indemnities and breach of data protection, security and confidentiality. Will liability for AI follow this path?
What does AI mean for Tech lawyers in 2026: (2) regulating AI
At the time of writing (end November 2025) AI regulation is in flux around the world. On 19 November 2025 the EU, struggling with a public image of over-regulation,[5] published its Digital Omnibus[6] packages to (per the European Commission’s press release[7]) “harmonise, clarify and simplify” the GDPR, delay and amend the AI Act[8] and revise the ePrivacy Directive to amend cookie rules. In particular, when enacted, the package will push back, likely beyond 2026, the AI Act’s rules on high-risk systems. The Digital Omnibus package also aims to consolidate reporting obligations under the GDPR, the NIS2 Directive[9] and the Digital Operational Resilience Act (DORA)[10], and simplify the EU Data Act[11] and the other data-related legislation it interacts with.
However, the Omnibus packages are in for a rocky ride with fierce opposition from digital rights groups, including Amnesty International[12] and Max Schrems’ NOYB[13], and unclear support from Member States and the EU’s other legislative bodies (including the European Parliament). The main views from the opposing camp are that the proposals (1) are a sign of bending to US and Big Tech pressure, as a backwards step for ongoing digital and data sovereignty concerns, (2) strip back many of the key protections, and (3) rather than “simplify”, add complexity, reopening settled positions, e.g. under the GDPR, that will require businesses to re-perform their GDPR and digital compliance programs.
Across the Atlantic, the next day, on 20 November 2025, the US president[14] sought to warn the US States off AI regulation at State level in favour of a single Federal Standard.[15] This is a redoubling of his previous (withdrawn) attempt to ban AI regulation for 10 years in his “big, beautiful bill” over the summer.
The UK, uncomfortably poised between a backpedalling EU and an unpredictable US, will be pressing ahead in 2026 with is own raft of data related legislation. This includes the Cyber Security and Resilience (Network and Information Systems) Bill (to replace the earlier legislation from 2018) and the Data (Use and Access) Act 2025 (which updates key parts of the Data Protection Act 2018). However, despite several promises pre and post election, for now, it remains unlikely that the UK will issue an AI Bill.
What does AI mean for lawyers in 2026: (3) copyright and IP
Stage 1 of the Data (Use and Access) Act includes a requirement for HMG to publish an impact assessment and a progress report on AI and copyright issues by March 2026. Whether the Government will move away from putting in place an opt-out model as its preferred policy option is likely to be contentious: it has already received over 10,000 responses to its initial consultation a year or so ago.
The UK Getty Images v Stability AI case of 4 November 2025[16] has shed little light on the substantive question whether AI models trained on copyright works infringe those copyright works, the case focusing mainly on jurisdictional issues and Getty’s secondary infringement claim failed, in essence because the Stability AI model used in the UK did not store or reproduce the original copyright works, despite the fact that it had been trained on them (outside the UK).
The UK High Court’s decision regarding the Stability AI model can be contrasted with the conclusions reached by the Munich Regional Court in GEMA v OpenAI, LLC and OpenAI Ireland Ltd[17], where the court found that 2 of OpenAI’s models infringed because song lyrics were reproducibly contained within the LLMs. These cases show that – from a legal perspective – each case will be assessed factually on its merits, meaning that there may be different positions in different jurisdictions and for different models.
2026 may see more progress on this substantive question as the US Getty / Stability AI litigation moves through the US system, particularly whether the fair dealing provisions of US copyright law will give model developers more comfort.[18]
It is anticipated that referral from Hungary to the European Court of Justice of the question of the extent of the text and data mining copyright exemption in the EU Digital Single Market directive[19] will not be decided before 2027.[20]
Other things to look out for in 2026
Although AI and regulation will continue to dominate the Tech landscape in 2026, the ripple effect of the potent combination of Cloud + AI will continue to impact more and more deeply regulated sectors like healthcare, financial services and professional services.
One day, possibly closer than we think, quantum computing will find real world use cases – probably cryptography and cyber security; and then how the law will deal with tiny assets that can be in more than one place at the same time will make the property characteristics of digital assets look simple.
[1] AI’s true impact will become apparent in the coming year, The Economist, 10 November 2025.
[2] v0.1_State_of_AI_in_Business_2025_Report.pdf, The Gen AI Divide: State of AI in Business, 2025, Massachusetts Institute of Technology, July 2025
[3] For an excellent chart, see the 2025 MAD (Machine Learning, Artificial Intelligence & Data) Landscape – mad2025.pdf. See also Bubble & Build: The 2025 MAD landscape, Matt Turck, FirstMark, 28 October 2025
[4] All chips in! Would a fall in AI-related asset valuations have financial stability consequences? | Bank of England, 24 October 2025. AI stocks in the basket included: Hyperscalers – Microsoft, Alphabet, Meta and Oracle; AI Chipmakers – Nvidia, AMD, Broadcom and Micron; AI Application Providers – Tesla, Uber and Salesforce; and certain utilities.
[5] E.g. How the EU botched its attempt to regulate AI, Barbara Moens and Melissa Heikkila, Financial Times, November 20, 2025
[6] Digital Omnibus on AI Regulation Proposal | Shaping Europe’s digital future and Digital Omnibus Regulation Proposal | Shaping Europe’s digital future
[7] https://ec.europa.eu/commission/presscorner/detail/en/ip_25_2718
[8] Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence and amending Regulations (EC) No 300/2008, (EU) No 167/2013, (EU) No 168/2013, (EU) 2018/858, (EU) 2018/1139 and (EU) 2019/2144 and Directives 2014/90/EU, (EU) 2016/797 and (EU) 2020/1828 (Artificial Intelligence Act)Text with EEA relevance.
[9] https://eur-lex.europa.eu/legal-content/EN/TXT/?uri=CELEX:02022L2555-20221227
[10] https://eur-lex.europa.eu/eli/reg/2022/2554/oj/eng
[11] Regulation (EU) 2023/2854 of the European Parliament and of the Council of 13 December 2023 on harmonised rules on fair access to and use of data and amending Regulation (EU) 2017/2394 and Directive (EU) 2020/1828 (Data Act)
[12] https://www.amnesty.org/en/latest/news/2025/11/eu-digital-omnibus-proposals-will-tear-apart-accountability-on-digital-rights/
[13] https://noyb.eu/en/open-letter-digital-omnibus-brings-deregulation-not-simplification
[14] Truth Details | Truth Social
[15] Donald Trump’s support for pro-AI proposal fuels Maga backlash, Joe Miller, Financial Times, November 20, 2025
[16] https://www.judiciary.uk/wp-content/uploads/2025/11/Getty-Images-v-Stability-AI.pdf
[17] Higher Regional Court of Munich 1, Case No. 42 0 14139/24, 11 November 2025. See press release at German_Court_OpenAI_Memory_Output_Infringe_Copyright_NOV25.pdf
[18] Getty Images (US), Inc. v. Stability AI, Ltd. et al 3:2025cv06891 | U.S. District Court for the Northern District of California | Justia
[19] DIRECTIVE (EU) 2019/ 790 OF THE EUROPEAN PARLIAMENT AND OF THE COUNCIL – of 17 April 2019 – on copyright and related rights in the Digital Single Market and amending Directives 96/ 9/ EC and 2001/ 29/ EC
[20] Like Company v Google Ireland Limited (Case C-250/25), referred on May 23 2025.
