AI-enabled cyberattacks like deepfakes, targeted phishing and data breaches are emerging threats for large and small organizations alike. Concerns around these threats are growing, with over 55% of survey respondents believing that genAI will ultimately give attackers a cyber advantage.45 Therefore, organizations need to stay abreast of developments in cybercriminal use of AI to pre- empt potential future attack vectors. For leaders to invest and innovate in AI with confidence, they also need to gain a comprehensive understanding of the cyber risks related to their adoption of AI. Key actions for leaders include: –Balance risk and reward: Regularly weigh AI’s operational benefits against potential cyber risks. –Invest in essential cybersecurity operations: Couple AI innovation investments with security investments to ensure security is embedded throughout the AI system’s life cycle. –Embed AI cyber risk into cross- organizational risk management: Involve multi-disciplinary teams to address the cyber risks from AI adoption effectively – either by adapting existing structures or creating new ones tailored to the unique challenges of AI. –Promote AI security-by-design and by- default: demand robust third-party risk management and use the organization’s purchasing power to promote AI security-by- design and by-default. –Engage with national and sector-specific standards: Stay informed about evolving AI regulations and how the specific local and regional AI context impacts business operations and risk. The use of AI in cyber controls can enhance defence strategies, making them more adaptive and efficient against evolving threats. Examples include improving threat intelligence mechanisms and automating vulnerability management. Cortex XSIAM by Palo Alto Networks, a cybersecurity firm, is a platform that offers advanced threat detection and automated responses by deploying machine learning and behavioural analytics. The platform aggregates and enriches data from various sources, allowing for real-time, adaptive threat management. To date, organizations that use the platform have been able to reduce incident investigations by 75% – and resolve 9 times more security issues than they previously did.46CASE STUDY 13 AI-enabled cybersecurity Telecom firm AT&T boosted its ability to detect fraud by up to 80% when it harnessed nearly 100 machine-learning models to process over 100 petabytes of data in real time. Thanks to these efforts, AT&T was also able to generate instant customer protection alerts and actionable insights across its call centres, stores and online channels.47CASE STUDY 14 Human-AI collaboration Deploying scalable AI strategies is dependent upon establishing a strong digital core – which consists of AI applications and digital platforms, a data and AI “backbone”, and physical and digital infrastructure (see Figure 4).48 Company-level enabler 3: Cybersecurity Company-level enabler 4: Digital core For leaders to invest and innovate in AI with confidence, they also need to gain an understanding of the cyber risks related to AI. AI Governance Alliance: Transformation of Industries in the Age of AI 18

AI in Action Beyond Experimentation to Transform Industry 2025