California’s largest trial courts are quietly experimenting with artificial intelligence to help decide your case, and current policies mean parties remain unaware when it happened. Los Angeles and Riverside County Superior Courts have contracted with an AI “clerk” called Learned Hand that drafts legal research, summarizes motions, and prepares tentative rulings for judges—all without mandatory disclosure to the people whose lives hang in the balance.
The AI assistant reshaping judicial chambers
Learned Hand combines models from major tech companies to function as a digital research attorney for overwhelmed judges.
Learned Hand, built by former Palantir executive Shlomo Klapper, combines large language models from Anthropic, OpenAI, and Google to function as a digital research attorney. The tool analyzes court filings, generates legal memos, and helps judges craft preliminary decisions. Los Angeles County signed a $314,000 contract allowing the AI to assist with civil, criminal, family, and probate cases, while Riverside’s smaller $10,000 pilot focuses on civil matters.
Six LA judges and their research attorneys are already using the system to streamline their crushing caseloads. The tool currently focuses on civil cases but contract language opens the door to much broader applications.
Criminal cases spark constitutional concerns
LA’s contract explicitly permits AI assistance on liberty-threatening motions that determine prison sentences.
The real controversy emerges around potential criminal applications. LA’s contract explicitly permits AI assistance on motions to suppress evidence and post-conviction relief petitions—decisions that determine whether someone goes to prison or gains freedom. Internal emails reveal courts have discussed expanding to Racial Justice Act petitions, where incarcerated people challenge convictions allegedly based on racial bias.
“When you’re dealing with someone’s liberty, the stakes couldn’t be higher,” warns LA District Attorney Nathan Hochman, who questions trusting AI with such consequential decisions. An anonymous LA judge called potential AI analysis of racial bias claims “outrageous,” arguing it would “dehumanize the whole process.” Public defender Elizabeth Lashley-Haynes describes using AI for such nuanced petitions as “highly problematic and bordering on unethical.”
Transparency loophole leaves litigants in the dark
Courts exploit a disclosure gap that only requires revelation when AI writes entire documents.
Courts exploit a crucial disclosure gap: they only must reveal AI use when artificial intelligence writes an entire document. Since Learned Hand assists rather than replaces human drafting, parties remain unaware when algorithms influence their case outcomes. Both courts declined to clarify whether litigants are being told that AI tools are being tested on their cases.
This pilot reflects our broader reckoning with algorithmic accountability in high-stakes decisions. Like Spotify curating your playlist, AI now helps curate justice—except unlike streaming recommendations, these algorithms operate without any transparency requirements. Until courts close this disclosure loophole, litigants are essentially participating in an undisclosed beta test every time they walk into these courtrooms.
