The Cloud Legal Project at Queen Mary University has done a research paper analysing the different terms of service from the major AI providers who offer their services in the UK and Europe. It's an outstandingly useful piece of work, but even better, they've made a website to help you navigate their findings.
If you're deciding which AI tool to deploy, this is such an important read. Importantly showing the difference between consumer and business terms, which can vary dramatically. This makes it even more important your team don't use their individual accounts on unapproved tools because this can really affect your contractual rights if problems arise down the line.
Really key things to think about in the approaches to terms which could impact your AI purchasing decision are:
- Who are you actually contracting with when you buy this AI tool — an Irish entity, an entity registered in a US State?
- Which law applies if you end up having a dispute and in which court do you have to bring a claim if it all goes wrong? This can really change the cost of a dispute and have you considered whether you have the legal expertise to assess the service under the law your AI provider says applies?
- What does the AI provider exclude liability for and how do they cap their liability? Spoiler, most caps are tiny…
- Does the acceptable use policy restrict what you can use the AI tool for? Lawyers beware, some of these terms say a "qualified professional" has to be in the loop to review all output, and they sometimes say you have to fully disclose use of the AI to your client. Are you complying with the terms of the underlying model your legaltech relies on?
This research doesn't extend to the intellectual property (IP) terms or the difference in terms relating to data and security (all topics our clients regularly ask us about), so add an assessment of these terms as well as the excellent work from Queen Mary to your procurement decision.
Link to the website:
Subscribe to receive more articles like this here.
