The 2026 Legal Playbook for Tech and AI businesses

A technology company creates legal risk at the same time as it creates value, and usually in the same act. The code that becomes the product is also the asset whose ownership has to be provable. The customer who validates the model is also the data controller whose terms have to be met. The playbook is organised around that, and around the observation that the legal work which is expensive to fix later is almost always cheap to do at the time.

The gaps that cause the most trouble

Two recur. The first is intellectual property that the company does not own. A founder who wrote code before incorporation, a contractor engaged without a written assignment, an agency that built the brand: in each case the rights stay where they were created unless they were assigned in writing. It surfaces during diligence, when a buyer or an investor asks for the chain of title, and it is at its most expensive at exactly that moment.

The second is a cap table and funding documents that create problems later. The Seed Enterprise Investment Scheme and the Enterprise Investment Scheme carry conditions on share class, on the use of funds and on the company's activities, and advance assurance is worth obtaining before terms are agreed rather than after. A convertible instrument signed quickly at pre-seed can cost several points of founder equity at the round that follows.

Stage by stage

The playbook works through five stages. Pre-incorporation and pre-seed, where the founder documents, vesting and leaver terms and IP assignments are put in place. Seed, where the legal stack has to support hiring, customers and investment at the same time. Series A and growth, where legal moves from documents to structure. International expansion, where the company structure, the employment position and the data transfer position all have to be worked through before entering a market. And exit readiness, which is described as something built over years rather than prepared in the three months before a sale.

Data, AI and compliance

There is a section on AI governance written for the questions a customer's procurement team will ask: where the training data came from and whether there are rights to use it, whether it contains personal data, whether customer data is used to train models, what testing has been done for bias and robustness before deployment, and what the company can say about all of this in writing. Most early-stage companies are not yet in scope of the heavier regimes, but enterprise customers and investors ask the questions well before regulators do.

The checklists

The closing pages are working checklists by area: founder documents, investment readiness, data protection and privacy, employment and people, and the AI questions above. They are the part most readers return to.

The playbook is a guide rather than advice on any particular company. If you would like the checklists worked through against your own position, our technology solicitors do that, and the intellectual property team handles assignments where the chain of title has a gap in it.