Posts tagged with "Defender for Cloud"

Google Threat Intelligence Group's 2 November 2026 LLMjacking report documents stolen cloud credentials driving fraudulent model charges past $100,000 in a single day and $1 million cumulatively, resold through underground "LLM Mixers" for $60 a month. The report names Azure Cognitive Services explicitly, so here is the six-control audit for Azure OpenAI: disable key auth, scan for leaks, log the data plane, cap TPM, alert on config drift, and decide on Defender for AI Services.