Telesign’s Responsible AI Policy

November 2023

Telesign is a provider of digital identity products and services which are designed to help businesses solve problems of user onboarding and account integrity, secure customer interactions, and reduce fraud. Telesign connects, protects, and defends businesses and customers by analyzing the digital interactions between them.

Digital identity verification has become a primary concern for many businesses, as banking, lending, entertainment, retail, and other sectors have gone digital.  Phone numbers are a unique identity anchor that help businesses solve identity verification challenges. This is where Telesign comes in, providing customers with analytics and insights about phone numbers through its suite of products, some of which are rooted in machine learning and artificial intelligence (AI).

AI delivers identity analytics with increasing efficiency and quality, enabling businesses to transact, communicate, and engage with their users free of fear. Our machine learning models are adept at uncovering hidden insights, predicting risk, and enabling customers to stay ahead of sophisticated attacks. However, while AI is capable of transformative business impact, we also acknowledge that it poses unique challenges and risks, including risks that may be difficult to anticipate. Telesign recognizes the need to define clear principles around the use of AI systems, and to be vigilant to any risks associated with their advancement.

With that in mind, Telesign has identified the following principles to guide the development and use of AI systems in our products and services. These principles also highlight the need for responsible usage from customers.

1. Human oversight – Telesign believes that AI systems should support customers in making better, more informed decisions about potential risks while maintaining an appropriate degree of human involvement and oversight. Customers should maintain control of their decision-making processes including, where possible, being able to challenge the results or reject the recommendations. This involves implementing an appropriate degree of human oversight to ensure the proper and responsible use of AI systems and reducing the possibility for adverse effects or unintended harms.
2. Technical robustness, accuracy, and reliability – AI systems should be designed with a preventative approach to risk, ensuring they perform as intended, operate accurately and reliably over time, and are resilient against adversarial attacks. This involves employing high-quality models, evaluating the provenance and utility of data sets, collaborating with dependable third-party vendors, and performing regular audits of AI systems. Our goal extends beyond operational excellence to foster a culture of trust, ensuring our technology meets customer needs and expectations.
3. Transparency – We are committed to being transparent about our use of AI systems, including our technology practices, data usage, and third-party vendors. We also strive to ensure that, wherever possible, our customers understand why AI systems produce recommendations and predictions so they can make informed decisions and take steps to address risks. This is why Telesign provides reason codes which help explain why a phone number has been given a particular risk score. We believe that AI systems should enable better and more equitable outcomes, and one of the ways to do that is to help customers understand the reasoning behind recommendations.
4. Fairness and non-discrimination – For AI systems to be considered responsible and trustworthy, they should lead to fair and non-discriminatory outcomes. Fairness entails that people are treated equitably and without discrimination. This does not necessarily mean identical outcomes for all people, but AI systems should not unfairly label certain groups as fraudulent based on non-relevant attributes or discriminate on the basis of fundamental and legally-protected characteristics. This involves continually reviewing AI systems for potential systemic bias and fixing any lapses as soon as they are discovered.
5. Security, privacy, and data governance – AI systems should guarantee security, privacy and data protection throughout the systems’ lifecycle. This is why Telesign’s AI systems prioritize security for customer data and are built in a way that embeds principles of data minimization and privacy by design. We also implement comprehensive data governance and management processes, including relevant design choices, rigorous data collection and preparation processes, and thorough assessments of training, validation, and testing data sets to ensure the data is relevant, representative, and error-free.
6. Accountability – We strive to be accountable for the development and use of AI systems. This includes ensuring that AI systems are designed with clear and traceable processes, maintaining comprehensive documentation, enabling the detection of potential errors or inaccuracies, and continually monitoring the performance of AI systems over time. This also includes implementing appropriate oversight and governance mechanisms. We are committed to learning from best practices and complying with evolving laws and regulations relating to the development and use of AI.

These principles reflect our broad commitment to ethical and responsible AI. We believe in a collaborative approach, where customers share in this commitment while utilizing our services. This includes managing AI outputs and recommendations responsibly, understanding the limitations of AI technology, actively preventing harmful outcomes, and staying in compliance with applicable laws and regulations. This approach ensures the mutual benefit of AI systems while mitigating unexpected outcomes and unintended harms.