Safe Offline, Safe Online: On AI and Cybersecurity

The sesion explored how AI can enhance cybersecurity measures while also presenting new challenges. The webinar featured David Alfred Hoffman from Duke University, Chelsea Smethurst from Microsoft, Nicolas Kourtellis from Telefonica Research, Lisa Green from Telstra, and Angel Melguizo from UNESCO, who highlighted the importance of capacity building, ethical principles, collaborative efforts, and practical implementation of AI-driven security solutions.

Prof. David A. Hoffman, CCLATAM think tank Founder and Professor of the Practice in the Sanford School of Public Policy, emphasized an interdisciplinary approach involving leadership from academia, the private sector, and government to tackle AI and cybersecurity challenges. He highlighted the critical role of trust and transparency in operations and technology usage, advocating for organizations to demonstrate reliability.

Hoffman outlined three main overlaps between AI and cybersecurity:

1. AI Improving Cybersecurity: AI tools can enhance threat detection, incident response, and vulnerability management. AI-driven Extended Detection and Response (XDR) tools, which analyze large datasets across distributed networks, can effectively counter ransomware attacks.
2. AI as a Cybersecurity Risk: AI can lower barriers to creating malware, enable more sophisticated phishing attacks by profiling individuals, automate impersonation, and impact distributed denial-of-service (DDoS) attacks. These risks necessitate protecting AI tools and training datasets from cyber threats.
3. Critical Infrastructure: The reliability of critical infrastructure, whether public, private, or a combination, depends on securing supply chains. AI can play a dual role in both enhancing and challenging these security measures.

Hoffman stressed the importance of equipping small and medium-sized enterprises (SMEs) within global supply chains with the necessary tools and knowledge to deploy these AI-driven cybersecurity measures, as they are often more vulnerable.

Chelsea Smethurst, Director for Digital Diplomacy at Microsoft, provided an overview of the cyber threat landscape and AI's role in enhancing cybersecurity. Smethurst shared impressive statistics, revealing that Microsoft defended against over 123,000 attacks in just 30 seconds, amounting to approximately 345 million attacks daily on their customers. This showcases the scale of cyber threats. By combining machine learning (ML) and natural language processing (NLP), Microsoft has developed AI security solutions that can predict and prevent cyber attacks more effectively. AI offers significant advantages in defensive threat intelligence, leveraging vast amounts of data, advanced infrastructure, and continuous innovation in building solutions. She highlighted how these AI-driven measures enable more effective prevention and response to cyberattacks, strengthening overall cybersecurity resilience.

Nicolas Kourtellis, Co-Director of Telefónica Research and Head of Systems AI Lab, detailed the company's approach to cybersecurity, emphasizing AI's role in the telecommunications industry. Telefónica leverages AI to optimize network performance, conduct predictive maintenance, enhance customer service, and analyze data. he explains how AI helps mitigate threats and identify anomalous behaviors from potential attackers, both inside or outside the network. Kourtellis highlighted AI's contributions to fraud detection, threat detection, and prevention by analyzing real-time data.

He highlighted Telefónica's commitment to ensuring compliance with privacy regulations, by implementing privacy-enhancing technologies, such as data anonymization and encryption. He highlighted Telefónica's efforts to ensure transparency and consent when using data, aiming to democratize AI deployment across various domains while prioritizing privacy, security, and fairness. Lisa Green, Data & AI Solutions Executive at Telstra, highlighted the dichotomy of excitement and fear when evaluating AI and cybersecurity. She noted the rapid evolution of technology, particularly generative AI, which presents both opportunities and challenges. While reduced barriers to entry for AI technologies enable small businesses to adopt these solutions, they often are not in the position to heavily invest in security controls or have the right skills in their workforce.

Telstra integrates cybersecurity into its responsible AI framework, making it a core component of all operations. In particular, the company focuses on confidentiality, data integrity, availability, and lately on abuse prevention, as generative AI introduces new risks like prompt injection attacks that can bypass safeguards and misuse AI technologies for harmful purposes. To stay ahead of threats, Telstra invests heavily in advanced security technologies and controls, ensuring continuous improvement. By using AI, Telstra is able to block an average of 280 million scam calls, e-SMS, and malicious emails monthly.

In terms of principles, Telstra priotitzes privacy protection, security reliability, and safety. She emphasized that while principles and policies are essential, embedding them into daily operations with clear guardrails and continuous improvement mechanisms is crucial.

She concluded by comparing cybersecurity to a team sport that requires a collaborative approach. Those collective efforts across industries are essential to effectively prevent cybercriminals and safeguard digital environments.