Global Data Breach Costs Drop to $4.44 Million, AI Security Gaps Widen

IBM’s new Cost of a Data Breach report reveals a striking trend: the rapid adoption of artificial intelligence is outpacing the development of security and governance needed to protect it. For the first time, the report includes a dedicated analysis of AI-related access controls and security frameworks—highlighting how quickly AI systems have become valuable and vulnerable targets for attackers.

A Growing Gap Between AI Use and Oversight

According to the study, 13 percent of organizations say they’ve experienced a security breach involving AI models or applications, while 8 percent are unsure whether such an incident occurred. Among those confirmed cases, 97 percent reported having no defined access controls for their AI systems. As a result, 60 percent of these breaches led to data exposure and 31 percent caused operational disruptions.

Suja Viswesan, VP of Security Product and Runtime at IBM, warned that attackers are already exploiting this oversight gap. She noted that the absence of basic controls is leaving sensitive data and model integrity at risk, and argued that AI security must be treated as “core infrastructure,” not an optional layer.

Still, organizations that have deployed AI and automation in their security operations saw notable benefits saving an average of $1.9 million per breach and reducing incident response times by 80 days.

Shadow AI: A Silent and Rising Threat

Sixty‑three percent of organizations affected by data breaches either lack an AI governance policy or are still drafting one. Even among those with policies in place, only 34 percent regularly assess unauthorized AI usage.

One in five organizations reported that a breach was directly triggered by Shadow AI unapproved or unsupervised AI tools used inside the company. Yet only 37 percent have formal policies to detect or manage such systems. Organizations with heavy Shadow AI usage paid, on average, $670,000 more per breach, and these incidents resulted in above‑average exposure of personal data and intellectual property.

AI tools are increasingly part of the attackers’ toolbox as well: 16 percent of all breaches involved adversaries using AI often in phishing campaigns or deepfake-based attacks.

Global Costs and Regional Trends

The average global cost of a data breach fell to The time to identify and contain breaches dropped to 241 days, down 17 days from last year. Organizations that discovered attacks internally saved roughly $900,000 compared to those notified by an external actor.

Healthcare remained the most expensive sector, with an average breach cost of More organizations refused to pay ransom in the past year (63 percent vs. 59 percent previously), though ransomware-related costs remain high, especially when attackers publicly leak stolen data.

Alarmingly, only 49 percent of organizations increased security spending after a breach, and less than half of those invested in AI‑driven security solutions.

Long-Term Fallout

Nearly every organization surveyed experienced operational disruption following a breach, with full recovery often taking more than 100 days. Despite slight year‑over‑year improvements, almost half of companies said they would raise prices for their products or services due to breach-related financial impact—nearly a third expecting price hikes exceeding 15 percent.