Sophos leverages behavioural AI and explainable tools to tackle cyber threats

SUNIL SHARMA
VP & MD - SALES (INDIA AND SAARC), SOPHOS

“Signature-based detection alone can no longer keep pace, especially in India’s rapidly growing digital ecosystem. At Sophos, we leverage behavioural models and anomaly detection that continuously learn normal baselines across networks, endpoints, and cloud systems. Deviations—such as lateral movement, privilege escalation, or unusual data flows—are flagged as potential zero-day or novel attacks. In India, AI- driven approaches are proving indispensable, detecting exploits faster than traditional tools. Once detection occurs, automated response orchestration isolates affected systems, throttles suspicious processes, or enforces microsegmentation, containing threats before human intervention. Analysts then validate, tune, and refine responses, enabling AI to learn from real-world feedback. Balancing AI-driven cybersecurity with privacy, ethics, and regulatory compliance is critical. We adopt privacy-by-design: collect minimal metadata or anonymized features, avoid storing raw personal data, and apply differential privacy or masking. Human-in-the-loop checks ensure expert review when decisions impact privacy. AI systems comply with India’s DPDPA 2023, mapping every decision flow to obligations, maintaining audit trails, and running algorithmic audits. Ethical guidelines such as MeitY’s Principles for Responsible AI embed fairness, accountability, transparency, and inclusivity.

No single organization can address threats alone. We collaborate with threat intelligence networks, academic institutions, government bodies, standards organizations, and open-source communities to share insights, validate models, and ensure interoperability. Transparency is reinforced through explainable AI, feedback loops, third-party audits, red-team tests, and stakeholder reports, ensuring AI decisions are understandable, accountable, and trustworthy while maintaining performance, fairness, and security.”