Cloud and AI bills looking a bit high? Your AI agents may have been let loose and run up huge spending costs (www.techradar.com)

Forcepoint has unveiled a critical issue regarding agentic AI systems that could significantly inflate enterprise cloud costs. As these AI agents operate with increasing complexity, a seemingly simple user request can trigger numerous downstream operations—termed 'unbound consumption'—leading to excessive use of compute resources, API calls, and tokens. This scenario poses a dual risk: not only can organizations face unexpected financial burdens from legitimate operations, but attackers could exploit these vulnerabilities to generate massive workloads using stolen credentials, ultimately passing the costs onto the enterprise. To mitigate these risks, Forcepoint emphasizes the need for enhanced visibility and control mechanisms within AI systems. Companies are advised to implement more granular budget limits at various levels, such as individual users and teams, and to enhance monitoring of cost attribution. Additionally, introducing agentic circuit breakers could prevent runaway resource consumption. These measures are becoming increasingly crucial as the intersection of AI complexity and operational costs demands closer scrutiny from both security and finance teams.