One Cable, 200 Gbit/s: Connecting Two GB10 Systems with ConnectX-7

What looks like a simple Ethernet cable is one of the most interesting parts of a two-node GB10 setup. A single QSFP112 cable can provide a dedicated 200 Gbit/s path for distributed inference, model training, MPI, an…...

$ EXECUTE_READ

Introducing DVAIB - Damn Vulnerable AI Bank

Can You Trick an AI Into Giving You Free Money?Introducing DVAIB - The World's Most Vulnerable AI BankWhat if there was a bank where the AI assistant was... a little too helpful? A bank where smooth talking might j…...

$ EXECUTE_READ

Multi-Turn Jailbreaks and “Psychological Manipulation” Attacks: Why Single-Turn Defenses Fail

Quick answerThe technique is Crescendo. It begins with individually benign questions, uses the model's earlier replies as context, and gradually plants and escalates concepts toward a harmful objective without askin…...

$ EXECUTE_READ

KV Cache Security for Long-Context LLM Inference

What Changed in 2026Short answer: Shared KV caches can expose prompt information when cache reuse crosses trust boundaries or unsafe distributed components are reachable. Separate cache reuse by trust group, use an unpr…...

$ EXECUTE_READ

Policy-as-Code for AI Workloads in Kubernetes: Kyverno/OPA Patterns for Model and Data Safety

2026 implementation noteKubernetes ValidatingAdmissionPolicy has been stable since Kubernetes v1.30 and is useful for CEL checks on admission objects, including labels, security contexts, resource limits, and approved a…...

$ EXECUTE_READ

Securing Agentic AI: Roadmap Part-10

Part 10. Implementation Roadmap10.0 Why you need a roadmap, not a random pile of botsYou now have:Agent patternsMulti agent topologiesHITL designsThreats and controlsIdentity, architecture, governanceGreat. Now the obvi…...

$ EXECUTE_READ

Securing Agentic AI: Governance Framework Part-9

Part 9. Governance Framework9.0 Why you need actual governance, not “vibes”At small scale, you can ship an agent, watch it in Prod, and fix things as they break.At enterprise scale, that same approach turns into:Nobody …...

$ EXECUTE_READ