Our AI baseline outlines the principles, practices and processes we rely on at Kyndryl to responsibly deploy and maintain artificial intelligence (AI) systems internally and for our enterprise customers.
"Kyndryl’s deep data expertise and our customer-first, vendor-neutral approach helps us focus on delivering business value while managing the ethical, legal and regulatory risks of AI. Our robust Al governance processes contribute to Kyndryl's reputation as a trusted Al partner.”
Michael Bradshaw,
Kyndryl Global Practice Leader Applications, Data and AI
As the world’s largest IT infrastructure services provider running mission-critical infrastructure and applications, Kyndryl has a unique perspective on AI’s practical implications. We believe that a principles-based approach to AI governance is the most effective way to leverage the significant business benefits offered by AI while simultaneously addressing regulatory and ethical requirements and associated legal risks.
Consequently, our responsible approach to AI is underpinned by our commitment to six fundamental principles:
- Transparency and explainability
- Detect and reduce bias
- Privacy and confidentiality
- Safety and security
- Accountability
- Education and training
Learn more about the Responsible AI at Kyndryl principles.
Grounded by our Responsible AI at Kyndryl principles, Kyndryl's AI governance framework establishes the guardrails that support the responsible adoption and use of AI within Kyndryl. This framework includes the Kyndryl AI Policy and AI Management System, plus additional guidance and resources designed to promote AI literacy and best practices.
AI Policy
The Kyndryl AI Policy applies to all Kyndryl employees and contractors and establishes the conditions for use of AI systems internally and in customer engagements.
AI Management System
Our AI Management System builds on existing policies and assessments with a risk-based review and approval process focused on issues relevant to the use of AI. Our AI System Inventory facilitates and accelerates the broader adoption of AI within Kyndryl by listing all AI systems approved for internal use as well as Kyndryl approved AI offerings and solutions for use in customer engagements.
AI at Kyndryl intranet portal
Our internal AI at Kyndryl portal outlines our AI governance framework and provides employees with access to Kyndryl AI policies, training, news, contract templates and other helpful materials that promote AI literacy and best practices.
Kyndryl has established several internal governance bodies to guide the implementation of our AI strategy.
Kyndryl AI Board
The Kyndryl AI Board has overall responsibility for the responsible adoption of AI within Kyndryl. Comprised of key members of Kyndryl’s senior leadership team, the AI Board meets regularly to review AI policy, strategy, partnerships and governance.
Kyndryl AI Governance Committee
The Kyndryl AI Governance Committee maintains the operational programs that support the AI Board’s mission and objectives and facilitates cross-functional exchange and decision-making on AI-related topics. The group includes representatives from the Kyndryl security team, CTO office, corporate strategy, legal and other functions, as well as representatives from our Global Alliance Partners.
Technical AI Review Boards
Comprised of Kyndryl distinguished engineers, our Technical AI Review Boards are charged with enabling Kyndryl technically through the development and promotion of professional standards across AI applications internally and externally.
Kyndryl has implemented a review process for all procured third-party AI systems to verify that the commercial solutions and technology stacks deployed by our vendors, suppliers and partners align with Kyndryl’s AI standards and policies.
Kyndryl’s AI terms, which form part of Kyndryl’s standard procurement terms, require all suppliers to inform Kyndryl if they use AI systems in their offerings and services. Kyndryl expects that all its suppliers implement adequate AI governance programs and take responsibility for compliance with laws, including with regards to the training of all underlying models.
Kyndryl has established a global AI literacy program to equip employees with the necessary knowledge and skills to navigate the evolving landscape of AI and to enable Kyndryls using AI systems to do so competently both internally and for customer engagements.
We provide access to numerous resources and training designed to help Kyndryls understand the opportunities and risks related to the use of AI systems.
These include:
- Annual required training: Each year, all Kyndryls complete extensive training on cybersecurity, privacy, data governance and AI. Training content is continuously adapted based on technological and regulatory changes.
- Additional education resources: Kyndryl offers additional training and education resources designed to strengthen Kyndryl’s understanding of the basic concepts of AI and the application of good data governance principles.
- Role-based training: We provide additional specialized training and certification programs for employees involved in the planning, development and use of AI systems. We work closely with our alliance partners to adapt the content of these programs to different practice areas, roles and levels of expertise.
The creation of Kyndryl’s Responsible AI at Kyndryl principles predates the arrival of legislation designed to regulate the use of AI, such as the EU AI Act. These principles set a high bar in terms of our ethical, privacy and security commitments and we continue to rely on them to guide our responsible use of AI.
We actively monitor the development of AI legislation around the globe and map any new legal requirements to our existing corporate policies and governance processes, where needed.
Where possible, Kyndryl refrains from creating country-specific processes and requirements. This approach helps ensure consistency, reduces fragmentation and facilitates compliance across jurisdictions, while still allowing for tailored adjustments where local laws require.
"AI is transforming how we work – from automation to agentic systems. But with greater autonomy comes a stronger need for trust, transparency, and robust governance. Kyndryl Consult works side-by-side with our customers to deploy AI with speed and confidence.”
Patrick Gormley,
Data Science and AI Consulting Lead
Kyndryl Bridge is Kyndryl’s Al-powered, open-integration digital business platform that Kyndryl consultants use to generate insights based on customer data and derive tailored recommendations aligned with their organization’s business goals. In support of our managed services customers, Kyndryl Bridge guides the creation of automations and performs operational IT tasks that include monitoring, anomaly detection and event correlation, enabling organizations to continuously innovate and achieve higher levels of operational maturity.
Kyndryl Bridge is powered by a mix of generic and dynamic machine learning (ML) models, as well as by pre-trained large language models (LLMs).
- Generic ML models process general infrastructure data pertaining to customer environments and yield contextual predictions. They are not trained using customer data.
- Dynamic ML models are trained to forecast metrics for a specific customer’s environment using general infrastructure data from that customer’s devices and applications. These models are not leveraged for any other customers and are deleted upon expiration or termination of a customer’s agreement.
- Large Language Models used by Kyndryl Bridge are obtained by Kyndryl pre-trained and are not further trained with customer data. Kyndryl-developed agentic AI workflows are also not customer-specific. All interactions with LLMs within Kyndryl are governed by strict guardrails.
The Kyndryl Agentic AI Framework enables enterprises to adopt, deploy and scale agentic AI-powered solutions with speed and confidence. Benefiting from Kyndryl's unique infrastructure-first perspective, the Framework was built with industry-standard encryption protocols, privacy-by-design principles and zero-trust security. Scalable automation and real-time decision-making coupled with human-in-the-loop oversight foster trust in AI-driven processes and provide assurances that agent actions are traceable, interpretable and continuously improved with human oversight.
All AI tools, systems and foundation models destined for use in Kyndryl service offerings and solutions, including Kyndryl Bridge and Kyndryl’s digital workplace solutions, undergo a comprehensive AI governance review process, providing customers with the reassurance that their data will be handled consistently in line with our AI policy and contractual obligations.
We provide transparency regarding our use of AI, including the guardrails we implement as part of the development process, while keeping the necessary freedom of action in our delivery.
Kyndryl employs rigorous development standards to ensure the quality and security of customer applications and solutions. We continually assess newly published standards and follow industry best practices for AI-related developments.
Code developed by AI to connect tools or further automate IT processes is tested with code verification software and further reviewed by a human before proceeding to testing in a closed environment.
Kyndryl’s AI use case review process is optimized to maintain agility in innovation and value delivery, while ensuring alignment with Kyndryl’s AI policies and principles.
The review process considers the business justification for each use case and assesses any underlying technical or operational risks. Based on the resulting risk classification, the appropriate regulatory authorities (e.g., the EU AI Office) are notified, as needed.
Kyndryl’s AI Management System is used to log and manage AI-related use cases from development through to deployment, including operational change management and ongoing monitoring.
If you have any questions related to privacy at Kyndryl, please contact us by using this form.
