Asia’s courts are drowning in paperwork. Can AI save them?
Firms across Asia say the bigger opportunity lies in automating the mundane tasks of local court systems
Add BT as a preferred source
ARTIFICIAL intelligence has made a bad name for lawyers over the years. Chatbots have generated fake case citations, fabricated legal arguments, and caused some lawyers – those who have relied on the tech too heavily – to issue apologies.
So why did the release of Anthropic’s new legal plugin in February help trigger what investors dubbed the “SaaS apocalypse,” wiping billions off software stocks globally in 2026?
The sell-off also hit Indian IT and software firms such as Infosys, Tata Consultancy Services, and Wipro, erasing nearly US$50 billion in market cap for the sector in February alone.
In another move earlier in May, Anthropic released 12 legal-specific plugins for Claude.
But the real opportunity for AI is not replacing lawyers entirely, says Sonam Chandwani, managing partner at KS Legal & Associates. Instead, it is “about reducing the massive amount of repetitive, process-driven work surrounding legal practice.”
Much of this is currently handled by junior associates, who are becoming increasingly concerned that these new plugins could have an impact on their jobs.
“The fear inside the profession is real,” says Denise Farmer, general manager of Asia Pacific at Clio. She warns that firms could begin reducing junior hiring “without thinking about where the next generation of senior practitioners will come from.”
“Of course, we worry. It’s our bread and butter,” says Malaysia-based lawyer Nazirah Mannan. She is concerned some areas of legal work could eventually become obsolete.
Go local or go West?
Research firm Technavio estimates that the generative AI legal market will expand by US$2.1 billion between 2024 and 2029, growing at more than 34 per cent annually.
SEE ALSO
Anthropic’s launch of the new plugins has intensified competition in a sector already flooded with money. US-based Harvey, for instance, raised US$200 million in March at an US$11 billion valuation.
Meanwhile, incumbents are rapidly reinventing themselves. Clio, founded in 2008 as a legal practice management platform, has repositioned itself as an AI-first legal operating system. It also acquired AI research company vLex in a US$1 billion deal in 2025.
Still, companies across Asia say the bigger opportunity lies in automating the mundane tasks of local court systems.
Whether these companies can build models that beat their Western counterparts remains to be seen, but several Indian AI firms, for example, are building tools tailored to local courts. These include courtroom transcription software, paperless case filing systems, and legal research platforms trained on regional laws and local languages.
The founders of India-based platform Adalat AI argue that frontier AI systems are poorly adapted for emerging markets. Claude or ChatGPT may summarise contracts well, but they are not trained for multilingual Indian courtrooms or regional accents.
This creates an opening for localised AI infrastructure, which could become a major competitive advantage as courts across Asia accelerate digitisation efforts.
The same approach is beginning to attract attention in other markets, including Indonesia and Malaysia where courts face similar challenges.
The AI battle inside courtrooms
“Justice is a logistics problem,” says Utkarsh Saxena, co-founder and CEO of Adalat AI. Courts in South and South-east Asia are often overburdened with paperwork and manual filing systems, which often cause administrative delays.
In many Indian district courts, judges still handwrite witness depositions because stenographers are unavailable, Saxena says. In some cases, courts have to rewrite records because other judges cannot read their colleagues’ handwriting.
To address this, Adalat AI is building courtroom infrastructure that includes live transcription systems and paperless filing tools tailored for Indian courts.
Adalat AI, a not-for-profit company, says it funds its activities through philanthropic grants, foundations and CSR funding.
Unlike many legal AI startups that rely on APIs from tech giants such as OpenAI or Anthropic, Adalat AI says it has built its own AI models in-house.
The platform trained its speech-to-text systems on Indian legal jargon, accents, and dialects, and it currently supports 15 Indian languages including Hindi, Malayalam, Odia and Kannada.
According to co-founder and CTO Arghya Bhattacharya, the company’s transcription models are “at least 15 per cent more accurate” than generic speech recognition systems used in law settings.
The company says its systems are already being used in thousands of courtrooms across 10 Indian states, while Kerala has mandated its transcription systems statewide.
The firm is also building a paperless court management system that digitises the entire lifecycle of a case, from filing to hearings. Lawyers can upload filings digitally, while court officials can review documents, flag defects and schedule hearings through a centralised dashboard.
Meanwhile, judges can access case files and case schedules from a single platform instead of navigating stacks of paperwork.
Adalat AI stores all judicial data on servers located within India, arguing that sensitive court records should not leave the country. It has also received interest from countries including Indonesia and is exploring partnerships across Asia and Africa, the founders tell Tech in Asia.
This is no surprise: The same localisation challenge extends across South-east Asia, where courts operate in diverse languages and legal systems that frontier AI models are not specifically trained on.
Malaysia-based lawyer Mannan says she is in discussions with two startups that are developing live courtroom transcription systems that could allow lawyers and judges to instantly access hearing records during proceedings.
Another India-based startup, Nyai, is building AI systems trained on millions of Indian judgments and statutes.
“At Nyai, if you put a research query onto the system and you get a 10-page output out of it, every paragraph will have a citation,” lawyer and co-founder Chinmay Bhosale says.
If the system does not have enough verified data to answer a query, it will simply say so instead of generating inaccurate or hallucinated responses, he says.
Beyond fake citations
Lawyers across Asia are increasingly using AI tools for research as well as drafting and compliance work despite ongoing concerns over hallucinations and inaccurate legal outputs.
Mannan says Claude is “95 per cent accurate” compared to rival AI systems for legal research, but she still manually verifies every response. OpenAI’s ChatGPT and some Chinese AI tools such as DeepSeek, she says, “always make up stories.”
Nyai’s Bhosale argues that most mainstream AI chatbots are designed to be “helpful” rather than legally reliable.
But the risks go beyond fake citations.
In a recent US federal court case involving a lawyer using Claude, the AI delivered legitimate legal citations but misrepresented the arguments contained within those cases.
Courts and regulators are increasingly paying attention to the risks of AI in the realm of law. The Supreme Court of India has warned against relying on AI-generated citations in legal proceedings.
Still, regulatory frameworks remain unclear across much of Asia. Mannan says Malaysian law still does not clearly prohibit non-lawyers from using AI tools for legal work.
India also lacks a dedicated AI policy, says India-based lawyer Salman Waris, who argues that regulation is inevitable as legal AI adoption accelerates.
“You cannot automate accountability,” Waris says. TECH IN ASIA
Decoding Asia newsletter: your guide to navigating Asia in a new global order. Sign up here to get Decoding Asia newsletter. Delivered to your inbox. Free.
Share with us your feedback on BT's products and services
