Infosys Open Sources Responsible AI Toolkit on its Topaz Suite

Infosys has launched an open-source Responsible AI Toolkit as part of its Topaz Responsible AI Suite, reinforcing its commitment to ethical AI adoption. The initiative aims to help businesses address AI-related risks while fostering a secure, fair, and transparent AI ecosystem.

The toolkit is built on Infosys’ AI3S framework (Scan, Shield, and Steer) and integrates advanced guardrails, including AI models and shielding algorithms to detect and mitigate security threats, privacy breaches, biased outputs, deepfakes, and other risks. 

It also enhances explainability by providing insights into AI-generated decisions without compromising performance. Designed for flexibility, the toolkit is customisable, compatible with diverse AI systems, and supports both cloud and on-premise deployments.

“As AI becomes central to driving enterprise growth, its ethical adoption is no longer optional,” said Balakrishna D. R. (Bali), executive VP, global services head, AI and industry verticals, Infosys. “By making the toolkit open source, we are fostering a collaborative ecosystem that addresses the complex challenges of AI bias, opacity, and security.”

Industry leaders have recognised the significance of the initiative. Joshua Bamford, head of Science, Technology and Innovation at the British High Commission, called it a “benchmark for technological excellence,” adding that it empowers enterprises, startups, and SMEs to adopt AI responsibly. 

Sunil Abraham, Public Policy Director – Data Economy and Emerging Tech at Meta, emphasised that open-source tools are crucial in ensuring AI safety, diversity, and economic opportunity.

The Government of India also welcomed Infosys’ move. Abhishek Singh, Additional Secretary at MeitY, noted that open-sourcing the toolkit would “help in mitigating bias in AI models and enhance security, privacy, and fairness in AI-based solutions.”

Infosys has been actively advancing Responsible AI initiatives. It was among the first companies to receive ISO 42001:2023 certification on AI management systems and is engaged in global AI policy discussions through organisations such as the NIST AI Safety Institute Consortium, WEF AIGA, AI Alliance, and Stanford HAI.