IBM (IBM) Status Update summary

AI and automation in cybersecurity

• AI is increasingly used for business process automation, threat detection, and customer service, with 29% of surveyed organizations applying AI at the intersection of security, especially for threat detection.

• Business leaders expect significant ROI from AI investments, but risk increases with greater AI integration, especially when managing AI pipelines internally versus SaaS consumption.

• Multiple business functions are co-sharing generative AI capabilities to build integrated processes and serve customers, with a focus on data-driven applications.

Cost of data breaches and AI impact

• The average cost of a data breach has risen, with healthcare experiencing the highest average at $9.77 million.

• There was a 22-23% increase in organizations facing regulatory fines over $50,000 due to breaches.

• Adoption of AI and automation in security operations centers increased by 10% year over year.

• AI and automation can reduce breach response time by 98 days and save organizations up to $2.2 million per breach.

Security risks and testing in AI environments

• Offensive security testing focuses on the entire AI ecosystem: MLSecOps pipelines, models, platforms, and GenAI applications.

• Risks include malicious model downloads, prompt injection, model bias, and expanded attack surfaces in cloud and internal environments.

• Security frameworks like OWASP and MITRE ATLAS are used, but practical gaps remain, especially in MLSecOps pipelines connected to sensitive data lakes.

• Red teams target data scientists and MLSecOps pipelines due to limited monitoring and logging in these environments.