Citadel AI Inc. (headquarters: Shibuya-ku, Tokyo; CEO: Hironori Kobayashi), a global leader in trustworthy AI technology, is pleased to announce that it has raised a ¥520 million Series A round from UTokyo Innovation Platform Co., Ltd (headquarters: Bunkyo-ku, Tokyo; president & CEO: Kosuke Ueda), Coral Capital (headquarters: Chiyoda-ku, Tokyo; founding partner and CEO: James Riney), ANRI (headquarters: Minato-ku, Tokyo; founding partner: Anri Samata), Suntory Holdings Limited (headquarters: Minato-ku, Tokyo; president & CEO: Takeshi Niinami), and Mitsubishi UFJ Capital Co.,Ltd. (headquarters: Chuo-ku, Tokyo; president: Yoichi Kasai).
The EU AI Act and global efforts to achieve “Trustworthy AI”
The rapid development of generative AI technology has accelerated concerns about AI reliability and trustworthiness in many countries around the world. On June 14, 2023, the world’s first AI legislation, the EU AI Act, was approved by the European Parliament. The regulation will require technical conformity assessments, risk management processes, and information disclosure for high-risk AI systems that affect human life, safety, and dignity.
Similar discussions are underway in the United States, Singapore, and other countries, and the AI Strategy Council has been established in Japan in parallel with the G7 discussions of 2023.
Ensuring “trustworthy AI” is now an urgent issue in today’s socio-economic landscape, and there is a strong demand for the development of international standards and frameworks, as well as the rapid implementation of technical solutions.
A versatile solution designed from experience deploying high-risk AI
Citadel AI provides unique technology for achieving trustworthy AI: Citadel Lens for automated reliability testing during the AI development stage, and Citadel Radar for automated monitoring during the operational stage. These two products instantly detect hidden risks in AI systems and visualize their causes in a human-readable format. Citadel AI’s technology is built from our team’s first-hand experience deploying high-risk AI systems at world-leading companies such as Google Brain, Waymo, Toyota, and more.
Conventionally, quality assurance of AI systems requires a manual, highly-skilled process with inconsistent and time-consuming evaluation techniques. Many evaluation techniques, such as explainability to interpret a model’s predictions, also require deep access to model internals, which are brittle and frequently break during model updates.
Our products utilize model-agnostic technology that can automatically evaluate, explain, and debug AI systems from the outside by testing input data and output predictions. This enables them to be applied across various models within a company as a unified AI reliability platform, and can continuously audit models during retraining to easily track evaluation history.
Global partnership with BSI on technical and legal compliance for high-risk AI
Citadel AI’s products have been adopted by BSI (British Standards Institution), a global leader in the field of international standards development and certification. Citadel AI was selected from among 54 companies worldwide after a year-long technical evaluation process. As a result, our technology will be deployed globally as a technical and regulatory assessment tool for high-risk AI in the fields of medical devices, automobiles, critical infrastructure, credit screening, and more.
Additionally, Citadel AI is a member of the AI Verify Foundation, a project run by the Infocomm Media Development Authority (IMDA) of the Singapore government to build “Trustworthy AI,” and included as one of 14 AI Assurance Techniques by the Center for Data Ethics and Innovation (CDEI) and the Department of Science, Innovation and Technology (DSIT) of the UK government.
This round of investment will enable us to expand the scope of our products to cover the rapidly evolving generative AI space, and to incorporate conformity assessment capabilities for new AI regulations and standards in various countries, further accelerating our business development in the global market.
Following our previous seed investment in Citadel AI, we are thrilled to lead this funding round as well. We met Kobayashi-san and Kenny-san during their company’s early stages and selected them for our incubation program, 1st Round. To be honest, we were initially a bit uncertain if societal needs would catch up with their cutting-edge problem setting.
However, the world and their company have changed dramatically in the past year. Nowadays, there is not a day when we don’t hear about AI regulations in the news, and Citadel AI’s product has been adopted as an AI evaluation system by BSI, one of the largest third-party certification bodies in Europe, where AI regulations are the most advanced. The company is rapidly transforming into a global startup attracting top-level engineers from around the world.
This is a one-in-a-thousand opportunity for a Japanese startup to play a significant role in shaping the global rules of AI regulation. We would like to seize this opportunity together with Citadel AI.UTokyo Innovation Platform Co., Ltd.
Takahiro Mizumoto, Chief Investment Officer/Partner
As AI technology advances and expands, there is a growing need for AI quality control and reliability assurance that differs from traditional software. This has become even more evident and concrete today than when Citadel AI was founded in 2020. The team members who have gathered under the leadership of CEO Kobayashi and CTO Kenny at Citadel AI are all sincere and talented, and are driven by social responsibility and technical challenges. We were impressed by the highly skilled members who are passionate about the mission of “trustworthy AI”, and decided to invest in Citadel AI.Coral Capital
Ken Nishimura, Partner
Our fund has decided to make an additional investment following the previous seed round. As generative AI rapidly gains popularity, we believe that the demand for Citadel AI’s services, which provide trustworthy AI, will only continue to grow. Additionally, Citadel AI’s technological expertise has been internationally recognized, including by BSI, one of the world’s leading certification bodies. We look forward to seeing them making significant leaps from Japan to the world and creating a global scale business!ANRI
Masahiro Sameshima, General Partner
Since our initial investment in April 2022, Suntory System Technology, our group’s specialized IT subsidiary, and Citadel AI have jointly developed a platform to ensure the stable operation of AI systems at Suntory. One example is the application of Citadel AI’s products to the “Pallet Recovery Prediction Model“. This enabled automation and efficiency improvements in the model’s operation, as well as a steady improvement in prediction accuracy.
Through this mutual collaboration, and Citadel AI’s work with BSI (British Standards Institution), we were able to confirm the quality, reliability, and verification capabilities of the company’s products. Moreover, with the approval of the world’s first AI regulation (EU AI Act) in the European Parliament in June, there is a growing demand for the company’s services amid the global trend of AI regulation strengthening.
This time, we have made an additional investment to strengthen our collaboration and relationship with Citadel AI and to support the company in accelerating its growth. Moving forward, we will continue to cooperate with Citadel AI to strengthen the operation and monitoring of AI systems within the Suntory Group and accelerate the group’s digital transformation (DX) initiatives.SUNTORY HOLDINGS LIMITED
Future Business Development Department
Mikio Aoki, Senior General Manager
Achieving “trustworthy AI” is an urgent issue in the modern world, and there is a strong demand for the development of international AI standards and frameworks, as well as the early implementation of global solutions. Citadel AI’s products are grounded in real-world experience, which can be applied across a wide range of AI technologies in a cross-cutting and versatile manner. These products are aligned with current trends and can be deployed in the global market. We empathize with Citadel AI’s strong sense of mission to achieve “trustworthy AI” and, as a result of our investment, we aim to contribute to the company’s business growth by leveraging our strengths as a member of MUFG.Mitsubishi UFJ Capital Co.,Ltd.
Yuichi Nishio, Investment Department No.3 /General Manager