IBM Report: Shadow AI Breaches Cost Companies $670K More, Reveals 97% Lack Proper Controls

 IBM’s 2025 Cost of a Data Breach Report, unveiled today in collaboration with the Ponemon Institute, discloses that breaches linked to employees’ unauthorized use of AI tools result in an average cost of $4.63 million for organizations. This amount is almost 16% higher than the global average of $4.44 million.

The study, which involved 3,470 interviews across 600 breached organizations, highlights the rapid growth of AI adoption surpassing security supervision. Although only 13% of organizations reported AI-related security incidents, 97% of those breached lacked adequate AI access controls. Additionally, 8% were unsure if they had been compromised through AI systems.

“The data underscores the existing gap between AI adoption and oversight, with threat actors beginning to exploit it,” stated Suja Viswesan, Vice President of Security and Runtime Products at IBM. “The report exposed a deficiency in basic access controls for AI systems, leaving highly sensitive data exposed and models susceptible to manipulation.”

Shadow AI, supply chains are the favorite attack vectors

The report reveals that 60% of AI-related security incidents led to compromised data, while 31% caused disruptions to an organization’s daily operations. Customers’ personally identifiable information (PII) was compromised in 65% of shadow AI incidents, significantly higher than the 53% global average. Governance emerges as one of AI security’s major weaknesses, with 63% of breached organizations either lacking AI governance policies or still in the process of developing them.

“Shadow AI is comparable to doping in the Tour de France; individuals seek an advantage without considering the long-term repercussions,” remarked Itamar Golan, CEO of Prompt Security, in an interview with VentureBeat. His company has identified over 12,000 AI applications and identifies 50 new ones daily.

VentureBeat continues to witness adversaries’ strategies outpacing current defenses against software and model supply chain attacks. Not surprisingly, the report identifies supply chains as the primary attack vector for AI security incidents, with 30% involving compromised apps, APIs, or plug-ins. As stated in the report: “Supply chain compromise was the most common cause of AI security incidents. Security incidents involving AI models and applications were varied, but one type clearly claimed the top ranking: supply chain compromise (30%), which includes compromised apps, APIs and plug-ins.”

Weaponized AI is proliferating  

Various forms of weaponized AI, including LLMs designed to enhance tradecraft, are rapidly advancing. Sixteen percent of breaches now involve attackers utilizing AI, primarily for AI-generated phishing (37%) and deepfake attacks (35%). Models like FraudGPT, GhostGPT, and DarkGPT, priced as low as $75 a month, are tailored for attack strategies such as phishing, exploit generation, code obfuscation, vulnerability scanning, and credit card validation.

The more specialized a particular LLM is, the higher the likelihood it can be directed to produce harmful outcomes. Cisco’s The State of AI Security Report reveals that finely-tuned LLMs are 22 times more likely to generate harmful outputs than base models.

“Adversaries are not only automating attacks with AI, but they are also using it to blend into normal network traffic, making detection more challenging,” explained Etay Maor, Chief Security Strategist at Cato Networks, in a recent conversation with VentureBeat. “The real challenge lies in the fact that AI-powered attacks are not isolated events; they represent a continuous cycle of reconnaissance, evasion, and adaptation.”

As highlighted by Shlomo Kramer, CEO of Cato Networks, in a recent VentureBeat interview: “There is a brief window during which companies can avoid being caught off guard with fragmented architectures. Attackers are moving faster than integration teams.”

Governance one of the weaknesses adversaries exploit

Among the 37% of organizations claiming to have AI governance policies, only 34% conduct regular audits for unsanctioned AI. Merely 22% perform adversarial testing on their AI models. DevSecOps emerges as the top factor in reducing breach costs, saving organizations an average of $227,192.

The report’s findings underscore how relegating governance to a lower priority impacts long-term security. “A majority of breached organizations (63%) either lack an AI governance policy or are in the process of developing one. Even when they do have a policy, less than half have an approval process for AI deployments, and 62% lack proper access controls on AI systems.”

Many organizations lack essential governance to mitigate AI-related risks, with 87% acknowledging the absence of policies or processes. Nearly two-thirds of breached companies neglect to audit their AI models regularly, and over three-quarters omit adversarial testing, leaving critical vulnerabilities exposed.

This pattern of delayed response to known vulnerabilities extends beyond AI governance to fundamental security practices. Chris Goettl, VP Product Management for Endpoint Security at Ivanti, emphasizes the shift in perspective: “What we currently call ‘patch management’ should more aptly be named exposure management—or how long is your organization willing to be exposed to a specific vulnerability?”

The $1.9M AI dividend: Why smart security pays off

Despite the increasing prevalence of weaponized AI, the report presents a hopeful outlook for combating adversaries’ evolving strategies. Organizations that fully embrace AI and automation save $1.9 million per breach and resolve incidents 80 days quicker. According to the report: “Security teams leveraging AI and automation extensively reduced their breach durations by 80 days and decreased their average breach costs by USD 1.9 million compared to organizations not utilizing these solutions.”

The stark contrast in outcomes is noteworthy. AI-driven organizations incur $3.62 million in breach costs, while non-AI organizations face costs of $5.52 million, resulting in a 52% cost differential. These teams detect breaches in 153 days, compared to 212 days for traditional approaches, and contain them in 51 days, as opposed to 72 days.

“AI tools excel at swiftly analyzing vast data sets across logs, endpoints, and network traffic, identifying subtle patterns early on,” Vineet Arora, CTO at WinWire, remarked. This capability revolutionizes security economics: while the global average breach cost stands at $4.44 million, extensive AI users operate 18% below that benchmark.

However, adoption continues to face challenges. Only 32% extensively utilize AI security, 40% deploy it in a limited capacity, and 28% do not use it at all. Established organizations evenly distribute AI across the security lifecycle, primarily focusing on prevention (30%), detection (29%), investigation (26%), and response (27%).

Daren Goeson, SVP Product Management at Ivanti, reinforces this: “AI-powered endpoint security tools can analyze vast amounts of data to detect anomalies and predict potential threats faster and more accurately than any human analyst.”

Security teams are not lagging behind; in fact, 77% match or exceed their company’s overall AI adoption. Among those investing post-breach, 45% opt for AI-driven solutions, emphasizing threat detection (36%), incident response planning (35%), and data security tools (31%).

The DevSecOps factor further enhances benefits, saving an additional $227,192, making it the top practice in reducing costs. When combined with AI’s impact, organizations can reduce breach costs by over $2 million, transforming security from a cost center to a competitive advantage.

Why U.S. cybersecurity costs hit record highs while the rest of the world saves millions

The cybersecurity landscape unveiled a notable paradox in 2024: while global breach costs decreased to $4.44 million, marking their first decline in five years, U.S. organizations witnessed a surge in exposure to an unprecedented $10.22 million per incident. This disparity signals a fundamental shift in how cyber risks manifest across geographical boundaries. Healthcare organizations bear the brunt, with an average breach cost of $7.42 million and resolution timelines extending to 279 days —five weeks longer than their counterparts in other industries.

The operational impact is equally severe: 86% of breached organizations report significant business disruptions, with three-quarters needing over 100 days to restore normal operations. Notably, post-breach security spending commitments dropped from 63% to just 49% year-over-year, indicating organizations are questioning the ROI of reactive security investments. Among those achieving full recovery, only 2% managed to restore operations within 50 days, while 26% required more than 150 days to recover. These metrics highlight a stark reality: while global organizations are enhancing their ability to contain breach costs, U.S. enterprises face a mounting crisis that traditional security investments alone cannot resolve. The widening gap necessitates a fundamental reassessment of cyber resilience strategies, particularly for healthcare providers navigating the intersection of heightened risk and prolonged recovery periods.

IBM’s report underscores why governance is so critical

“Gen AI has lowered the barrier to entry for cybercriminals. … Even low‑sophistication attackers can leverage GenAI to write phishing scripts, analyze vulnerabilities, and launch attacks with minimal effort,” observes CrowdStrike CEO and founder George Kurtz.

Mike Riemer, Field CISO at Ivanti, offers a positive outlook: “For years, attackers have been leveraging AI to their advantage. However, 2025 will mark a turning point as defenders begin to harness the full potential of AI for cybersecurity purposes.”

IBM’s report provides actionable insights for organizations:

1. Implement AI governance immediately – With only 45% having approval processes for AI deployments





2. Enhance visibility into shadow AI – Regular audits are crucial when 20% experience breaches from unauthorized AI





3. Accelerate adoption of security AI – The $1.9 million savings justify proactive deployment

As the report concludes: “Organizations must ensure chief information security officers (CISOs), chief revenue officers (CROs), chief compliance officers (CCOs), and their teams collaborate regularly. Investing in integrated security and governance software and processes to bring these cross-functional stakeholders together can help organizations automatically discover and govern shadow AI.”

As adversaries weaponize AI and employees create shadow tools for productivity, surviving organizations will embrace the benefits of AI while meticulously managing its risks. In this evolving landscape, where machines engage in rapid battles beyond human capabilities, governance transcends mere compliance to become a necessity for survival.