New LatticeFlow Platform Evaluates Governance of AI Systems

(CONNECT) LatticeFlow has released AI GO! According to a statement, this is the first platform to deliver in-depth technical assessments across any AI system. This new development from the company builds on COMPL-AI, the first framework for evaluating generative AI models under the EU AI law. It was developed by LatticeFlow, the Swiss Federal Institute of Technology in Zurich (ETH) and the Bulgarian Institute for Computer Science, Artificial Intelligence and Technology (INSAIT) and was welcomed by the EU Commission as a first step towards implementing the EU AI law.

"AI governance is only functional and scalable if it is based on technical insights," Petar Tsankov, CEO and co-founder of LatticeFlow, is quoted as saying. "With AI GO!, companies gain the deep insights they need to address AI risks, prove compliance and accelerate innovation with confidence."

According to the company, which was founded in 2020 as a spin-off from ETH, the new platform combines precise technical evaluation with automation and provides companies with a uniform overview across departments and use cases. It enables user teams to use ready-to-use frameworks for AI risks such as the EU AI Act or ISO and NIST standards or to integrate their own customized AI risk frameworks. In addition, AI GO! can be extended across modalities and applications, from generative AI to large language models and chatbots to computer vision and tabular models.

"The EU AI law is a turning point for the industry," says Tsankov. "It requires concrete technical proof that AI systems are safe and trustworthy. With AI GO! we help companies to directly meet compliance requirements and at the same time drive innovation and value creation forward." ce/mm