Legal translations often need to be done quickly, but above all they must be accurate and secure. For years, legal teams have relied on highly trained human translators to provide this level of reliability in translated contracts, patents, and regulatory filings. But sometimes that’s not fast enough, volumes are just too great, and humans are prone to error. So where does AI fit in? We met with Jae Kim, the co-founder of Bering Lab, to fill us in.
Jae Kim and Seong Moon founded Bering Lab in 2012. At that time, they provided legal translation services in the traditional model, using subject-matter experts to handle high-stakes documents and it became Korea’s largest legal translation agency.
But in 2018, Google’s AI-powered neural machine translation (NMT) breakthrough signaled a massive shift. AI was getting faster—but was it good enough for legal work? It certainly held promise for an industry that needed high volumes to be handled quickly.
Jae realized that adapting to AI wasn’t optional—it was imperative for the future of legal translation. In 2020, his team began developing BeringAI+, an AI translation software platform plus human review process designed to bridge the gap between AI efficiency and expert oversight.
Here, we talk with Jae about how we’re reshaping legal translation for the future.
🚀 Artificial Intelligence: An Existential Opportunity for the Translation Industry
📤 Interviewer: Jae, before we dive into AI and legal translation, tell us a little about your background. What led you to this industry?
💬 Jae Kim: I’ve always been drawn to technology and business. I studied computer science and business at Penn, and growing up in Silicon Valley, I was surrounded by tech innovation. My dad was in the tech business also. I started my career in investment banking and private equity, but after a few years, I wanted to return to tech and entrepreneurship. That led me back to Korea, where I co-founded and scaled a Korean beauty ecommerce company that became the largest at the time in Southeast Asia. Around the same time, I also started a legal and financial translation agency, which eventually became the largest provider in Korea.
📤 Interviewer: Jae, let’s talk about the moment AI truly started reshaping the translation industry. What was the turning point?
💬 Jae Kim: It was around 2018, when Google’s neural machine translation (NMT) breakthrough happened. I remember the first time we tested an NMT model—we expected something slightly better than Google Translate, but what we saw shocked us. The model was capturing legal terminology, industry-specific phrases, and even subtle language distinctions better than we had ever seen from previous AI tools. It was a revelation for us.
📤 Interviewer: So, you saw right away that AI would change the industry?
💬 Jae Kim: Yes, but at the same time it was a bit unsettling. Until then, like many in the translation industry, we believed that legal translations could only be done by human experts. Suddenly, we were looking at an AI model producing results that, in some cases, outperformed generalist human translators who lacked deep legal expertise.
📤 Interviewer: Did that make you see AI as a threat or an opportunity?
💬 Jae Kim: Honestly, I think everyone in the industry saw it as a little of both at first. But very quickly, we realized that adapting to AI wasn’t just a competitive advantage—it was imperative for survival, an existential opportunity. We knew that if we didn’t evolve, we would fall behind. And we realized, too, that we had access to vast amounts of non-proprietary legal-specific data that could be used to train these engines for even better performance. That’s when we started developing Bering Lab’s AI-powered translation approach.
📤 Interviewer: What made you confident that AI had a place in high-stakes translation?
💬 Jae Kim: The demand for faster, more scalable translation was growing rapidly, especially in highly regulated industries. Companies were dealing with hundreds, sometimes thousands, of pages of legal documents that required translation, but traditional methods were too slow and too expensive and could not scale. AI offered speed, but the challenge was making sure it was accurate, reliable, and secure—especially for legal use cases where even a small error could lead to serious consequences.
📤 Interviewer: So that’s where Bering Lab comes in—taking advantage of AI’s efficiency while improving its accuracy for regulated industries?
💬 Jae Kim: Exactly. We knew AI translations models by themselves weren’t enough. That’s why we focused on developing domain-specific AI models, trained on high-quality legal datasets, and integrating expert human review into the workflow. This hybrid approach gives clients the speed of AI without sacrificing the accuracy and accountability that legal translation demands.
🔎 The Challenges of Specialized Translation and AI’s Role
📤 Interviewer: Jae, AI tools like ChatGPT have made huge advancements in language processing. Why aren’t they enough for legal, patent, or life sciences translation?
💬 Jae Kim: General AI translation models are impressive, but they’re built for broad, everyday language, not for specialized, high-stakes content. Legal, patent, and life sciences documents require technical accuracy, jurisdiction-specific terminology, and contextual understanding that off-the-shelf AI simply can’t guarantee.
Legal translation isn’t just about getting words from one language to another—it’s about preserving the meaning, intent, and legal weight of every document. A single mistranslated legal term can change the meaning of a contract or lead to compliance issues. You wouldn’t trust ChatGPT to write a legal brief—why trust it to translate one?
📤 Interviewer: And I imagine security adds another layer of complexity?
💬 Jae Kim: Absolutely.
Security is a major issue. Many AI tools process data in the cloud, raising concerns about confidentiality and compliance. That’s why Bering Lab offers on-premise solutions for clients who can’t risk sensitive documents being exposed to third-party AI systems.
📤 Interviewer: So, you use AI to support human translators rather than replacing them?
💬 Jae Kim: Exactly. AI eliminates obstacles and handles repetitive tasks, while human expertise guarantees the final output meets legal and regulatory standards. It’s not a question of AI or human translators—it’s about using both in the right way.
💡Bering Lab’s AI-Powered Solutions
📤 Interviewer: Jae, we’ve talked about the challenges of legal translation and why AI alone isn’t enough. Let’s shift to solutions—what exactly does Bering Lab offer?
💬 Jae Kim: At Bering Lab, we have developed our own domain-specific AI models trained on high-quality multilingual legal and scientific datasets, which allows those models to handle specialized terminology with far greater accuracy than general AI.
From there, we help clients adapt the models with their own proprietary data, so translations reflect their internal language, preferred terminology, and compliance standards. This level of customization makes AI translation far more reliable for legal and enterprise use.
📤 Interviewer: That covers translation, but that’s only one part of providing translated documents to legal entities. Legal and regulatory documents involve a lot of other manual work. How do you address that?
💬 Jae Kim: That’s where our mini AI models come in. Legal translation isn’t just about converting text—it involves preserving document structure, correcting OCR errors, and managing terminology, all of which can be time-consuming and prone to inconsistencies when done manually.
We’ve developed AI-powered tools to remove inefficiencies in the following areas:
- Formatting: Maintains the original structure and layout of documents, reducing the need for manual reformatting after translation.
- OCR Correction: Fixes inaccuracies from scanned documents, improving text extraction for legal use.
- Glossary & Terminology: Builds custom glossaries so specialized terms remain consistent across translations.
📤 Interviewer: In some industries and for some content types, raw machine translation can be “good enough.” Since you’re working in regulated industries, is there always a human involved?
💬 Jae Kim: Yes, always recommend it, whether it is provided by us or done by our client. We also know even the best AI isn’t infallible. A legal professional always verifies accuracy, compliance, and liability protection before anything is finalized. Our clients need to know they can trust the output, and that’s why human oversight is built into the process.
One concept that’s extremely important in legal translation is accountability. Many industries, including legal, have requirements for sworn translations and certified translations, and that means the idea is that there is a human expert who signs off on the translation and stands behind it. That’s not going away.
🎯 Looking Toward the Future: How Bering Lab is Making Impossible Translations Possible
📤 Interviewer: Where do you see the future of AI in translation?
💬 Jae Kim: AI will continue to get faster, smarter, and more sophisticated, but human expertise will be needed at various stages of the process for the foreseeable future. Professionals will be involved at the start, in customizing and training the model and developing prompts, and at the end, with post-editing. Plus, there’s that need for accountability that I mentioned earlier. It’s hard to hold an AI accountable in the same way as a human.
Future AI development will be shaped by the companies developing the technology like us, the companies purchasing it, and the people using it. The companies that embrace AI as a tool, not a replacement, will be the ones that thrive.
At Bering Lab, we’re focused on expanding globally to help more clients translate higher volumes at quality and speed while continuously improving our AI-based approach. We recently added 30 languages to our platform, and we’re scaling into patents and life sciences alongside legal translation.
Our ultimate goal is to make high-accuracy, high-security translation faster, smarter, and more accessible—without compromising the trust that legal teams rely on.
📤 Interviewer: For businesses looking to modernize their translation processes, where should they start?
💬 Jae Kim: If you’re in a regulated industry where accuracy, security, and speed all matter, let’s talk. Whether you need AI-powered legal, patent or life sciences translation, secure on-premise solutions, or scalable document processing, we’re ready to help.
Bering Lab works with businesses in regulated industries to implement AI-driven solutions that fit real-world legal and compliance needs. Get in touch at sales@beringlab.com to see how our technology can support your team.
Is your company struggling with legal document translations? Does using AI in translation make operational and financial sense, but it doesn’t provide the results you need or expect?
BeringAI+ takes the delays and uncertainty out of the process by being faster, more accurate, and cost-effective. 👉 See the difference for yourself and get a FREE QUOTE today
