Theta Lake has announced a raft of updates to its AI-native platform, including a new certification, a fresh patent, new risk detection capabilities, and the launch of an educational series aimed at financial services professionals.
The platform has achieved the Cloud Security Alliance (CSA) STAR for AI Level II certification. Combined with Theta Lake’s existing ISO/IEC 42001 accreditation, the new certification draws on the CSA’s globally recognised Security, Trust, Assurance & Risk (STAR) for AI framework and its AI Control Matrix (AICM) to provide macro governance, granular security, and continuous validation. The certification is designed to tackle concerns around bias mitigation, model risk management, algorithmic explainability, and the privacy of training data.
Alongside the certification, the company has secured a new patent covering its topographic techniques for high-dimensional data analysis, selection, and labelling. The patent supports more efficient classifier engineering with improved performance and consistency. Theta Lake now holds more than 18 patents across AI technology and security and governance implementations.
Building on these foundations, Theta Lake has released several new risk detection classifiers targeting AI-based communications. These include a tool to detect when users try to manipulate AI-generated summaries by instructing them to omit or alter topics discussed — a behaviour that could compromise compliance or legal analysis. Another classifier identifies documents and attachments shared with AI assistants during interactions, while a third targets indirect prompt injections — instructions concealed within emails, documents, or chats that attempt to manipulate AI tools into bypassing safety guardrails or gaining access to sensitive information.
Theta Lake provides an AI-native compliance platform designed for regulated industries, offering tools for monitoring, forensic investigation, and governance of digital communications and AI interactions. Read the full article.
