IBM announced on July 29 that the share of organizations planning to increase cybersecurity spending rose to 85% in May, up from 64% in the period from March 2025 to February 2026, after learning about the cyber capabilities of frontier AI models.
The company attributed the increase to awareness of the advanced capabilities of frontier AI. About 75% of organizations said threats of this kind led them to rethink how they deploy AI agents in their security operations. More than half use agents for threat detection and containment, but only 18% apply them to vulnerability management.
"When organizations have a prolonged gap between discovery and remediation, that imbalance shows up directly in breach costs," said Suja Viswesan, vice president of IBM Security Software. "The priority now is to eliminate that lag — embedding remediation into development workflows, protecting identity at runtime, and fixing risks at the speed attackers already move."
The survey, conducted from March 2025 to February 2026, showed that AI-enabled attacks accounted for about 25% of malicious breaches reported in the period. That share grew 56% compared with the previous year. These breaches cost victims an average of US$ 6 million, about US$ 1 million above the US$ 4.99 million average. In the financial services sector, the average cost was US$ 6.3 million. Most attacks involved deepfake or AI-powered malware.
Context
OpenAI announced on July 21 that its models caused a security incident during tests of cyber capabilities, classifying the case as "an unprecedented cyber incident." On August 5, models from Meta and Anthropic breached other companies in cybersecurity tests. In April, the FBI reported that it received 22,364 complaints of internet crimes referencing AI in 2025.
