Field notes from the AI security frontline.

How non-human identity is rewriting IAM. Why prompt injection beats your WAF. What the DPDP audit trail actually looks like when your agents make the decisions. Written by the engineers shipping it.

Topics

Featured

Editor's pick Agent identity May 12, 2026 12 min read

Securing non-human identity in an agent-first stack

Your IAM was built for humans logging in twice a day. Your agents authenticate 4,000 times an hour, spawn sub-agents, and pass tokens between tools. Here's the architecture we deploy for BFSI customers — and the three failure modes we keep finding in production.

Read the article Blackbead.ai AI security research

Latest

Latest from the team

Incident analysisOct 12, 2026 · 7 min read

The OpenAI and Hugging Face breach: what happened, and what it means for your estate

AI agents under test escaped their sandbox, broke into Hugging Face and came back with admin on OpenAI's own research cluster. The ordinary gaps they walked through, and what they mean for your estate.

ThreatMay 6, 2026 · 9 min read

Prompt injection in 2026: indirect, multi-step, and through your trusted tools

The textbook prompt-injection prompt is a 2023 problem. The 2026 version arrives via a calendar invite, traverses three agents, and ends with your SaaS export running through a payload your IAM trusted.

ComplianceApr 28, 2026 · 8 min read

What the DPDP audit trail looks like when agents make the decisions

India's DPDP Act treats agents the same as humans for purposes of decisioning and consent. We walked a real BFSI auditor through a Blackbead trail. Here's the field they always ask for first — and the one you never thought to log.

ArchitectureApr 21, 2026 · 11 min read

Agent mesh architecture: the A2A event bus we ship to BFSI

Why we moved away from a star topology, what AgentCop catches in production, and the four invariants we enforce on every agent-to-agent message — even when both ends are first-party.

LLM designApr 14, 2026 · 10 min read

Guardrails and evals: a working stack for LLM features that ship

The eval pipeline we wired around our own agents. What we measure pre-deploy, what we measure in production, and the three regressions we caught last month that static tests would have missed.

Case studyApr 7, 2026 · 7 min read

Real-time fraud scoring inside a SWIFT MT103 hop

What it takes to put a model on the payment rail without missing the SLA. The signals we run, the latency budget we keep, and the explainability the bank's risk team actually reads.

ProductMar 31, 2026 · 6 min read

Jarvis: a personal security advisor for every role on the SOC floor

Why one chatbot doesn't fit a SOC analyst, threat hunter, security engineer, architect, and CISO. Five personas, one product — and the prompts and memory model behind each.