More than two-thirds of legal professionals say the cost of checking AI outputs now outweighs the benefits the technology was supposed to deliver, according to a survey of 850 senior lawyers and in-house counsel across the US, UK, Australia, and the Middle East. The report, "The Intelligence Gap: Why AI in Legal Isn't Delivering - and What Needs to Change," from alternative legal services provider Morae, found that 67% of respondents believe the burden of human verification has become a net negative, turning AI from a time-saver into a repair function.
Only 33% of respondents trust the results of AI-assisted legal work. Nearly half (48%) flagged poor accuracy, hallucinations, and errors as a constraint on effective AI use, while security and confidentiality concerns followed at 41%.
Human verification becomes repair work
The survey reveals that 41% of legal professionals check every AI output, and another 48% review outputs for sensitive content. Almost half (48%) said humans always or often materially change AI outputs before they can be used. Morae said this suggests human intervention is functioning as a repair mechanism, not a safeguard.
Half of respondents expressed concern about potential liability if AI-assisted work leads to errors or adverse outcomes. Reputational risk from AI errors worried 62% of those surveyed.
Data quality undermines AI reliability
Poor data quality is a root cause of AI underperformance, the report found. More than a third (37%) rated the structure or formatting of their data as poor or very poor, and a similar share (33%) said data consistency was poor. Only 26% of legal leaders feel confident their organization's legal information is AI-ready.
Many organizations lack a clear inventory of their data. Some 44% said they use multiple internal systems without central tracking, 39% lack a comprehensive file inventory, and another 39% said legacy data has never been properly classified or deleted. More than three-quarters of respondents said data quality issues undermine AI outcomes regardless of how good the AI systems are. Improving data quality is essential for reliable AI for Legal applications.
In-house teams demand transparency from outside counsel
The report highlights a transparency gap between in-house legal departments and their external law firms. Forty-one percent of in-house teams said transparency around external counsel's AI usage is limited or completely absent.
Eight in ten general counsel are concerned about how external firms use their data in AI systems. Fifty-five percent worry their data is being used to train AI models without their knowledge, and 51% are concerned data is being shared with AI vendors without consent.
Why this matters for legal professionals
The findings underscore that adopting AI without robust data governance and verification workflows can increase risk and cost rather than reduce them. Legal teams that focus on data readiness and develop skills to critically evaluate AI outputs will be better positioned to turn the technology into a genuine asset. Structured training, such as the AI Learning Path for Paralegals, can help teams build the verification and oversight skills needed to make AI tools a net positive. Firms must also address internal data quality and demand clear AI usage policies from external counsel to close the transparency gap.
Your membership also unlocks:
