The accelerated rhythm of adoption of artificial intelligence (AI) in companies is exposing a worrying imbalance regarding data management safety. According to the latest report Cost of Data Breach Report 2025 Published by IBM, the integration of AI in corporate processes has become a priority objective for cybercriminals, while protection measures do not advance at the same pace.
This gap between innovation and security generates critical vulnerabilities, especially in sectors such as finance, health and logistics, where the use of AI to automate decisions and process large volumes of data is increasingly common. The report also reveals that organizations that implement IA without a robust cybersecurity strategy face greater risks of leaks, loss of sensitive data and reputational damage.
Experts warn that it is urgent to adopt proactive approaches, such as the use of defensive, continuous audits and specialized training, to mitigate the impact of possible attacks and guarantee a secure and sustainable digital transformation.
Gaps in the AI era
The report, prepared by the Ponmon Institute and sponsored by IBM, reveals that 13% of organizations have already suffered gaps that compromise AI models or applications, while an additional 8% does not even know if they have been a victim. In addition, 97% of the affected companies did not have specific access controls for AI, which generated that 60% of these incidents implied data exposure and 31% caused operational interruptions.
The integration of AI in corporate processes has become a priority objective for cybercriminals, while data management protection measures do not advance at the same pace
“The report reveals a clear gap between the speed with which the AI and the supervision measures that accompany it are being adopted, a lagoon that the attackers are already taking advantage of,” warns Suja Viswesan, vice president of security products and execution environments in IBM. “The lack of basic access controls leaves sensitive data exposed and models vulnerable to manipulation. If AI will be integrated structurally into business processes, your safety must also be structural. What is at stake is not just money: it is trust, transparency and control.”
In spite of these risks, the report shows that organizations that apply the IA and intensive cybersecurity automation manage to reduce the average cost of a gap by 1.9 million dollars and shorten their life cycle in 80 days.
Governance and hidden risks
The document emphasizes that 63% of the affected organizations lacked AI governance policies, and only a third of those that do implement them perform periodic audits. In addition, the unauthorized use of AI, known as Shadow AIcaused one in five companies to suffer incidents, increasing costs by 15% compared to the global average.
The sophistication of cyber attacks has also grown: 16% of the attacks analyzed used AI to execute phishing campaigns or generate Deepfakes, complicating detection and mitigation.
Economic and operational impact
Globally, the average cost of a gap was reduced for the first time in five years, standing at 4.44 million dollars, although in the US, it reached a historical record of 10.22 million. The health sector remains the most affected, with an average cost of 7.42 million dollars and a recovery time of 279 days.
The report also indicates that ransomware attacks remain expensive, with an average cost of 5.08 million, and that half of the companies plan to increase prices after suffering a gap, with increases of up to 15%.
