While people are increasingly open to letting artificial intelligence handle simple legal paperwork, very few would trust AI with deeply personal matters like divorce, according to a new survey.
Commissioned by legal intelligence platform Robin AI, the study of over 4,100 people across the US and UK found that only 17% would be comfortable using AI to assist with divorce proceedings. By contrast, nearly half (47%) would trust AI to help write a will and 49% would trust it to review a rental agreement – suggesting that while AI is gaining ground in low-risk tasks, it faces major resistance in emotionally complex legal areas.
The findings are part of a wider crisis of confidence in both the traditional legal industry and emerging legal technology. Just one in ten people surveyed said they fully trust law firms, while only 4% would trust an AI “robot lawyer” acting alone. Instead, the overwhelming majority prefer either a traditional lawyer (69%) or a lawyer supported by AI tools (27%). Richard Robinson, CEO and Founder of Robin AI, said:
“The legal industry needs serious reform. People want faster, cheaper legal help, but not at the expense of human judgment. That’s why we’re focused on building AI to work alongside lawyers, not replace them.”
The public clearly distinguishes between administrative legal tasks and more sensitive disputes. AI is seen as suitable for less personal matters like:
- Reviewing rental agreements (49%)
- Writing a will (47%)
- Challenging parking tickets (46%)
- However, support plummets for more complex legal issues:
- Divorce – 17%
- Redundancy disputes – 17%
- Criminal defence – 11%
A strong majority (61%) said they would not trust AI to assist with criminal law at all, highlighting a sharp drop-off in confidence as the stakes rise. Only 27% believe an AI-powered paralegal could match the performance of a qualified human lawyer.
Cost remains a powerful motivator. When asked what would most encourage them to use AI for legal services, respondents cited lower cost (47%), 24/7 availability (35%), and faster resolution (35%). On average, people said they would require a 57% discount to choose AI over a human lawyer.
Beyond AI, the survey exposed significant frustration with the legal system itself. Nearly two-thirds (64%) described it as “expensive,” 34% said it’s “for rich people,” and 30% found it “intimidating.” Only 10% believe legal services are truly accessible to everyone, with most saying that quality legal help remains the preserve of those who can afford it. Robinson said:
“The current legal system wasn’t built for speed, transparency, or accessibility – it was built for those who could afford to wait, navigate the jargon, and pay the bill. When 90% of people say legal services aren’t truly accessible, that’s not just a flaw – it’s a systemic failure.”
He believes human-led AI can offer a rare chance to rebuild a fairer legal system, if developed with the right safeguards.
While overall trends were similar across the US and UK, some key differences emerged. UK respondents were more likely to support mandatory AI safety or compliance training for lawyers (82% versus 76% in the US). Meanwhile, trust in large corporate law firms was slightly stronger in the US, with 13% trusting Big Law “a great deal” compared to just 8% in the UK.
UK participants were also more likely to describe the legal system as “elitist” (23% versus 18% in the US).
Despite their reservations, many respondents see clear benefits in AI if used responsibly. Legal AI was associated with being more affordable (47%), less biased (38%), more accurate (24%), and more transparent (21%).
