DynamoFL Evolves to Dynamo AI: Empowering Enterprises with Secure AI

Over the past two years, DynamoFL has enabled some of the world’s largest enterprises to productionize AI in compliance with data and AI regulations. Our journey began with a focus on high impact use-cases where we deployed our differentially private federated learning platform to enable regulatory compliance, while maximizing model performance. 

In just the past year, we witnessed enterprises embark on a new era of AI, which has brought a wave of emerging compliance, security, and privacy challenges and major regulatory changes. Providing a truly comprehensive compliance solution for our enterprise customers required that we extend our offering beyond privacy-preserving training to include comprehensive model red-teaming, monitoring, and guardrailing solutions as well. 

For example, DynamoFL highlighted critical privacy and security vulnerabilities in Generative AI solutions, which could result in PII data leakage and systematic misuse. While privacy-preserving model training has an important role to play in the responsible development of Generative AI, productionizing Generative AI now requires also addressing downstream compliance challenges during the following red-teaming, remediation, and deployment phases. 

Additionally, as we deployed production-level federated learning, compliance with data export laws required that we also integrate a suite of model red-teaming and stress-testing capabilities to demonstrate robust defense against data leakage and reverse engineering attacks that can be performed on federated models. Importantly, this entire process needed to be fully auditable and explainable to regulators. 

To enable our customers to adopt an end-to-end solution to enable secure and compliant AI, we are thrilled to announce our renaming to Dynamo AI. Dynamo AI is an end-to-end platform for productionizing secure and compliant AI for enterprises. The Dynamo AI product suite spans the entire AI productionisation workflow, including:

1. AI system evaluations and red-teaming
2. AI system enhancement and risk remediation
3. AI system deployment and real-time observability. 

Specifically Dynamo AI’s new product suite includes the following product modules: 

DynamoEval: Offering 20+ privacy evaluations, security assessments, and hallucination tests, along with automated red-teaming documentation to map and control AI risks in alignment with emerging regulatory guidance and frameworks like OWASP Top-10 and MITRE ATLAS. 

DynamoEnhance: Providing comprehensive techniques to mitigate the risks identified through DynamoEval. This includes differential privacy, federated learning, PII redaction, advanced model safety alignment techniques, and RAG Hallucination remediation capabilities, along with many others.

DynamoGuard: Offering highly customizable and robust AI guardrails to defend against emerging threats, while ensuring the highest standards of security and efficiency. DynamoGuard includes a comprehensive LLM observability platform to audit every LLM interaction for compliance violations and misuse. 

While expanding the scope of our impact to new AI domains beyond differentially private federated learning, Dynamo AI will continue to boost its investment in its federated learning offering (now a part of DynamoEnhance), while simultaneously working to introduce exciting capabilities to Dynamo AI’s federated learning customers, such as data leakage testing for federated models, advanced differential privacy integrations, and industry-leading optimizations to run AI models privately on edge devices.

We’re thrilled to continue our work with enterprises as Dynamo AI and invite you to join us in shaping the future of safe and secure AI.

‍