AI Access Rights Become a Financial Risk for Executive Management
What This Is About
Google Threat Intelligence Group and Mandiant document in a report dated 2 November 2026 a growing form of attack called “LLMjacking”: criminals obtain access to companies’ cloud accounts using stolen or exposed credentials and run AI services on the company’s bill. The incidents affected organisations in North and South America as well as the Asia-Pacific region from financial services, retail, technology and transport. Unlike the gradual resource consumption from cryptojacking, fraudulent invoices in LLMjacking can spike sharply within hours. According to Google, costs in individual cases exceeded USD 100,000 on a single day; at some organisations they accumulated to as much as USD 1 million before access was suspended.
Why This Matters for the Board and Executive Management
These cases demonstrate that access rights to AI services directly impact costs and thus the profit and loss statement. In extreme cases, this can be existential for a small organisation. The most common entry point in the cases examined was exposed or stolen cloud identity tokens such as IAM credentials, service account keys, GitHub access tokens and active session tokens. This makes identity and access management a governance issue. For the board, this means in my view: it should be able to rely on controls that identify and contain such financial surprises early, and executive management should be able to make access to AI services and unusual cost spikes transparent before they turn into six-figure or higher losses. Clearly defined spending limits also help constrain potential damage.
Questions for Executive Management
- Do our systems detect unusual cost spikes in cloud and AI services within hours, or do we only learn about them from the invoice?
- Have we set appropriate spending limits, so that we have a fixed cost ceiling for potential losses and know what these limits are?
- How do we ensure that access keys to our cloud and AI services do not lie exposed in code repositories or configuration files, and that active session tokens cannot be intercepted?
Sources
- The Rise of LLMJacking: How Threat Actors Monetize Compromised Cloud and Model Resources (Primary source)
- Cyberkriminelle kapern LLM-Zugänge und treiben Cloud-Rechnungen hoch (SwissCybersecurity.net, 06.11.2026)
- Cyberkriminelle kapern LLM-Zugänge und treiben Cloud-Rechnungen hoch (Netzwoche, 06.11.2026)
- Google Warns LLMjacking Attacks Are Costing Some Enterprises Up to $1 Million in Cloud Charges (The Hacker News, 03.11.2026)