AI:
Consumers expect a lawyer to oversee the information and advice that artificial intelligence (AI) provides, major new research has found.
This is one of five red lines uncovered by the Legal Services Board (LSB) from its probe into the standards and safeguards that consumers want to see put in place around the technology, to protect them if things go wrong.
The LSB research found strong consumer support for legal AI in general, with around three-quarters of people believing AI would make legal services easier to use, more affordable and more accessible.
This was driven in part by the growing use in everyday life of platforms like ChatGPT and Gemini.
But there was real concern about the lack of standards and safeguards, with standards written specifically for consumer-facing legal AI tools “almost entirely absent”.
The LSB identified five red lines that consumers are not prepared to cross:
- AI tools must provide minimum guarantees for accuracy;
- AI tools must not take consequential action without prior informed consent;
- There must be human oversight and transparency when AI tools are used;
- There must be accountability for harm caused by AI tools; and
- AI tools must not compromise safety.
In two new reports – AI in Legal Services [1]by MEL Research and Existing standards for AI-powered business-to-consumer lawtech [2] by ox.ai – the LSB recognised an emerging problem.
“While these expectations are widely shared, the research highlights a significant gap between these expectations and what may be feasible in practice,” said the first report.
“Specifically, there are tensions between maintaining affordability and meeting expectations around accuracy, accountability and oversight. This ‘expectation-reality gap’ presents a key challenge for regulators and providers.”
Richard Orpin, the LSB’s chief executive, explained: “More and more people are turning to AI for help with legal problems, whether that’s a housing dispute, a problem at work, or a debt they cannot manage.
“For many people who cannot afford legal advice, these tools could be genuinely transformative.
“We commissioned this research because very little existed on how these tools are working for consumers. What we found is a gap between what people reasonably expect and the protections currently in place.
“Consumers are open to AI’s potential, but they expect basic safeguards, and right now those safeguards are largely absent.”
The research found: “In the absence of clear, accessible and AI-specific standards, lawtech providers have arrived at their own working frameworks through a combination of adjacent regulatory requirements, professional instinct, and trial and error.
“For B2C lawtech tools that operate without the involvement of legal professionals, this assumption creates a structural gap: the regulatory obligations that would ordinarily protect the consumer do not apply.”
People accepted that there would be differences between free and paid-for legal AI tools, in terms of quality and user experience, but they rejected the notion that low-cost or free services should expose them to more risk or harm.
The LSB said: “Safety is seen as a non-negotiable baseline, not a premium feature.”
Consumers also made it clear that any privacy violations attached to legal AI, or actions being taken without consent, serious inaccuracies, or not being able to speak to a human, are all “unacceptable”.
To address these issues, the MEL research said there was likely to be a role for voluntary action by providers, sectorwide initiatives – such as quality marks – and regulation.
Members of the LSB’s public panel “broadly” accepted the idea of regulation in order to meet their expectations.
“However, regulation might well require legislative change in order to apply. There are risks of unintended consequences (e.g. stifling innovation and increasing costs), and trade-offs must be made where expectations are in tension with one another,” MEL said.
Alongside the research, the oversight regulator has set out its AI plan for 2026-27 [3], with the aim of unlocking growth by enabling “safe AI innovation”.
It will do this by providing both “coordination across the legal services regulatory system” and “overarching regulatory certainty and clarity on AI adoption within which the frontline regulators can operate”.
This research forms part of the third pillar – building trust among consumers and small businesses.
Mr Orpin said: “As we develop our priorities for the coming three years, we will focus our efforts where regulation can have the sharpest impact, working with regulators, government, innovators, and consumer organisations to consider what targeted and proportionate action could look like, to protect consumers and support economic growth in the sector.”