WBN Responsible AI Policy

WBN is committed to being open about how AI is used in our products and what role it plays in the content and intelligence our audiences receive. We believe that transparency about AI is not just good ethics — it is a precondition of trust.

• Product-level disclosure. Every WBN product that incorporates AI includes a clear description of how AI is used in that product. That description is accessible to users and written in plain language.
• Content-level disclosure. Where AI has played a material role in generating or substantially editing a specific piece of content, that content carries a clear disclosure label. We do not present AI-generated content as independently authored human work.
• Methodology availability. The general methodology behind our AI-powered intelligence indexes, signal rankings, and analytics tools is documented and available to subscribers and partners on request. We do not require proprietary trade secrets to be disclosed, but we do not hide the basic logic of our systems behind unnecessary opacity.
• Policy updates. When WBN deploys a significant new AI capability, we update this Policy before or at the time of deployment. Our audiences are not surprised by AI capabilities they did not know existed.
• Honest about limitations. We communicate clearly when our AI systems have known limitations — including in accuracy, coverage, or context — so that users can apply appropriate judgment when relying on AI-generated outputs.

WBN uses AI to assist in the creation of editorial content in specific, defined, and disclosed contexts. The use of AI in content creation is not a shortcut to lower editorial standards — it is a tool to extend our editorial capacity while maintaining those standards in full.

WBN uses AI to personalise the content experience for registered users — surfacing news, intelligence, and features most relevant to their professional interests, industry focus, and usage patterns. Personalisation is designed to increase relevance, not to create filter bubbles, reinforce bias, or maximise engagement at the expense of editorial breadth.

AI systems require data to function. How that data is collected, used, and protected in the context of our AI products is governed by our Privacy Policy. This section addresses the specific intersection of data and AI at WBN.

• Data minimisation. Our AI systems are designed to use the minimum personal data necessary to achieve their purpose. We do not collect broader data than we need in order to make AI systems more capable.
• No sale of AI-derived profiles. WBN does not sell profiles, inferences, or other data derived from AI processing of user behaviour to third parties. AI-derived insights about users remain within WBN's operational environment.
• Training data standards. Where WBN uses user interaction data to improve or train its AI models, this is done on the legal basis described in our Privacy Policy, using data that has been appropriately prepared and anonymised where required. We do not use sensitive personal data to train AI models without explicit consent.
• Third-party AI training. WBN does not provide user data to third-party AI providers for model training purposes without explicit user consent and a clearly documented legal basis.
• Retention limits. Data used in AI processing is subject to the same retention limits described in our Privacy Policy. AI systems do not retain personal data beyond what is necessary for their defined function.

AI systems learn from data, and data reflects the world — including its inequities, its historical patterns of discrimination, and its gaps in representation. WBN takes the risk of AI bias seriously and addresses it as an ongoing operational responsibility, not a one-time technical review.

The security and safety of WBN's AI systems is treated with the same rigour as the security of our infrastructure and the safety of our editorial content. AI systems introduce specific security risks — including adversarial manipulation, prompt injection, and model extraction — that we address through dedicated controls.

• Adversarial robustness. WBN tests its AI systems against attempts to manipulate their outputs through adversarial inputs — including prompt injection attacks on language model systems. Systems that are found to be vulnerable to manipulation are not deployed in contexts where such attacks could produce harmful outcomes.
• Access controls. Access to WBN's AI infrastructure, model weights, and training data is restricted to personnel with a defined operational need. Access is logged and reviewed.
• Output monitoring. AI outputs in deployed systems are monitored on an ongoing basis for anomalous patterns that may indicate system compromise, model drift, or unexpected behaviour.
• Incident response. WBN maintains defined response procedures for AI-related security incidents, including rapid shutdown capability for affected systems and communication protocols for notifying affected users where required.
• No autonomous escalation. WBN's AI systems do not have the capability to autonomously expand their own permissions, access new data sources, or take actions outside their defined operational scope without human authorisation.

WBN's AI capabilities include both proprietary systems built by our own teams and capabilities provided by third-party AI providers. In both cases, the standards set out in this Policy apply. Where a third-party AI provider cannot meet our standards, we do not use their technology in ways that affect our audiences.

Responsible AI practice requires more than good intentions — it requires structure, accountability, and defined processes that make ethical practice systematic rather than occasional. WBN maintains a governance framework for its AI systems that is integrated into its operational and editorial leadership.