One of legal's hottest startups is helping lawyers finally answer: Is the AI's work any good? (www.businessinsider.com)

Crosby, a pioneering legal tech startup, has introduced Redline Bench, a novel tool designed to assess the performance of AI models in legal tasks, beginning with contract review. This initiative addresses the critical challenge of defining quality in legal outputs, which is inherently subjective compared to more quantifiable fields like coding. By creating a benchmark that evaluates how closely AI redlines align with the priorities of experienced lawyers, Crosby aims to enhance trust in AI tools within the legal community, where the stakes are high amidst significant investments aiming to automate legal processes. Using input from senior lawyers to establish weighted criteria for important contract changes, Redline Bench employs a panel of judges to score AI-generated edits against these benchmarks. The initial results rank ChatGPT 5.5 as the top performer, highlighting the progress AI is making in legal tasks. This benchmarking approach not only offers transparency and credibility to AI's capabilities in law but also encourages ongoing improvements, making it easier for legal professionals to embrace these technologies. As Crosby and other firms develop standardized evaluation methods, the legal sector could see transformative efficiencies and cost reductions driven by AI.