We Wanted Smarter Legal Tech, but Instead Got an Expensive Dependency

Key Takeaways

• Law firms' AI spend rose 9.7% in 2025, but efficiency gains lag.
• Only 6% of firms pass AI savings to clients; 34% raise rates.
• eDiscovery AI review costs fell to $0.11‑$0.50 per document.
• Verification and compliance add hidden costs, eroding net productivity.
• EU AI Act compliance could cost $2‑$15 million per firm.

Pulse Analysis

The legal sector’s AI boom mirrors a broader enterprise trend: hefty budgets outpacing tangible returns. Firms poured billions into generative tools, expecting faster document review and leaner operations, yet surveys show adoption has plateaued at 79% and real‑world efficiency gains remain modest. The mismatch stems from entrenched hourly billing structures that capture AI‑driven speed without translating it into lower client fees, creating a paradox where technology fuels profitability for firms but inflates costs for end‑users.

A deeper issue lies in the verification tax that accompanies AI output. Studies across industries reveal that up to 37% of purported time savings are consumed by reviewing and correcting machine‑generated work. In legal practice, where precision is non‑negotiable, this hidden labor can nullify the theoretical advantages of generative tools. Moreover, the eDiscovery niche demonstrates both promise and peril: per‑document AI review costs have dropped dramatically, yet the lack of robust quality‑control frameworks raises the risk of privileged‑information breaches and spoliation, potentially outweighing cost benefits.

Regulatory headwinds add another layer of complexity. The EU AI Act, effective August 2026, classifies many legal‑tech solutions as high‑risk, imposing compliance expenditures ranging from $2 million to $15 million per organization. Coupled with U.S. bar association guidance mandating human verification of AI citations, firms must budget for extensive risk‑management programs. These emerging costs, combined with stagnant productivity, suggest the industry must rethink measurement metrics, adopt alternative fee arrangements, and prioritize governance to unlock genuine value from legal AI investments.