Cybersecurity platforms are rapidly adopting generative and agentic AI, but the hidden shift to token-based consumption is creating unpredictable costs that can overwhelm SOC budgets during active incidents. The article explains how ML, GenAI, and agentic AI differ in cost behavior and warns that vendors like Anthropic’s Claude and OpenAI-style APIs can turn deep investigations into expensive, variable operations. #Anthropic #Claude #PaloAltoNetworks #GPT-5.5
Keypoints
- Agentic AI can rapidly investigate security incidents, but it consumes tokens continuously.
- Token-based pricing makes cybersecurity spending unpredictable and difficult to budget.
- Large investigations can burn millions of tokens in a short time.
- Organizations may face tradeoffs between paying overages, throttling analysis, or falling back to manual triage.
- On-premises deployment can help control AI costs better than cloud-based metered usage.
Read More: https://www.securityweek.com/the-ai-token-costs-that-can-break-cybersecurity/