Security researchers are reporting a sharp increase in LLM-jacking attacks, where cybercriminals compromise corporate AI accounts and servers to fuel a new wave of cyber crime. These incidents target expensive AI resources, often exploiting weak access controls or unpatched vulnerabilities in large language model deployments. The Financial Times highlights that attackers are not only stealing computational power but also using hijacked systems to train malicious models or execute fraudulent transactions.

Organizations relying on cloud-based AI infrastructure are particularly vulnerable, especially when multi-factor authentication is absent or misconfigured. Experts urge companies to implement strict identity verification, continuous monitoring, and automated anomaly detection to mitigate risks. As AI adoption accelerates across finance, healthcare, and logistics, the attack surface expands, making robust cybersecurity frameworks essential.

The trend signals a shift from traditional data breaches to resource-based exploitation, demanding new defensive strategies tailored to AI-specific threats.