AI legal issues

Continuing disputes are likely to occur in relation to the rights of those whose data has been ingested to produce the large language model on which the AI has occurred.  The recent judgment in Getty Images v Stability has shown that copyright claims in this area may be difficult, but that is unlikely to be last word.  As the decision has shown, trade mark disputes are still a risk for AI developers.

Whilst AI can certainly offer extraordinary outputs, defining the required specific performance for an AI solution can be difficult.  And in the absence of clear performance requirements, disputes can occur.  

AI can be prone to a range of errors from hallucinations to producing defamatory or discriminatory outputs.  The potential liability for these errors is only just beginning to be explored.

Much of the substance of AI output can constitute personal data and this can bring a host of issues under data protection law including the potential requirement to disclose the basis of the algorithm being used to process data.  For an AI solution, that can be highly complex.

Within the EU there is now a bespoke AI regulatory framework in the form of the EU AI Act.  That does not exist yet in the UK but AI solutions may be subject to existing and sector specific regulatory regimes and this is an ever changing landscape.

Although tied to regulatory approaches particularly across the UK, EU and US, companies of all shapes and sizes are increasingly expected to abide by a set of ethical standards in the development, creation and use of AI tools. Reputationally, ethics is increasingly determining consumer behaviour. But implementing a set of ethical principles requires care to avoid imposing undue restraint on commercial activities, whilst limiting exposure to  PR and legal risk in the event of a breach.