IBM: AI Cyberattacks Up 56%, Breaches Cost $6.04M

AI-driven cyberattacks are on the rise, with more than one in four organizations reporting malicious incidents linked to artificial intelligence. IBM states that these AI-enabled attacks increased by 56 percent year-on-year. The average cost of a data breach involving AI is now $6.04 million. This is about $1 million more than breaches where AI was not a factor. Attackers are using generative AI to automate reconnaissance and craft convincing social engineering content. This shortens the time needed to exploit vulnerabilities. Deepfake impersonation is the most common AI-driven tactic, making up 45 percent of incidents. AI-generated malware accounts for 19 percent, while AI-assisted phishing is 17 percent. Attackers are abandoning human speed for machine speed, according to IBM. Businesses' own AI deployments are also being targeted. 21 percent of organizations experienced a security incident involving an AI model or application. This is up from 13 percent a year earlier. The most costly incidents were model inversion, averaging $6.07 million, and prompt injection attacks, averaging $5.89 million. This means organizations must prioritize protecting AI models and improving detection of AI-assisted deception to reduce financial and reputational risks.