Featured
Trust Infrastructure: The Real AI Bottleneck
AI capability is here. The bottleneck is trust. Every transformative technology follows the same pattern, and AI is stuck in phase one.
Read articleWhat Is a Creed?
A creed is a machine-readable constitution that governs AI behavior: more specific than a system prompt, more flexible than a filter, and portable across models.
Read articleAlignment Isn't Solved. Here's What's Missing.
Some researchers are declaring victory on AI alignment. They're confusing one kind of progress with the whole problem. The structural gap (whose values, enforced how,...
Read articleAutonomous Response, Zero Governance
A security engineer built an AI triage system that auto-remediates production alerts. The tool is excellent. The governance is absent.
Read articleDashboards Are Not Governance (Yet)
Tracking how developers use AI tools is observability. Governing what AI tools actually do requires a different set of capabilities entirely.
Read articleGuardian: Constitutional AI That Runs on Your Machine
Guardian is a free, open-source safety evaluator. Install it, point it at text, get a verdict. No cloud, no account, no API costs.
Read articlePreparing for AI Compliance: What You Actually Need
The EU AI Act is in force. NIST published its AI RMF. Most teams aren't ready. Here's what the regulations require and how to meet...
Read articlePsychosecurity: When AI Gets Manipulated
Prompt injection is a known threat. Psychosecurity addresses a different class: sustained adversarial influence that degrades AI behavioral integrity over time.
Read articleRLHF Has a Thermodynamic Limit
RLHF tries to maintain a low-entropy behavioral surface on a high-entropy capability space. The second law sets a deadline. What comes after suppression?
Read articleThe Perfect Impersonation Problem
An email agent so good nobody can tell it apart from its owner. That is the problem, not the achievement.
Read articleWhen Your AI Tool Forgets to Ask
A solo salesperson built an AI call recorder that works perfectly as software and fails completely as a governed system. A personal creed would have...
Read articleWhy Consent Frameworks Will Not Work for Home Robots
Embodied AI is arriving faster than the privacy frameworks designed to govern it. BIPA and CCPA assume bilateral transactions between a user and a service....
Read articleWhy Guardrails Aren't Enough
Most AI safety products filter inputs and outputs. This catches obvious violations and misses everything else. Constitutional AI takes a different approach.
Read articleYour Agent Can Call Stripe Directly. Should It Make Ethical Decisions Directly?
Generic tool-calling is becoming a commodity. Constitutional AI governance is where the real stakes sit, and why MCP matters more for ethics than for API...
Read articleYour Agent Has Permissions. Does It Have Principles?
Autonomous AI agents need more than access control. They need constitutional governance: values that guide judgment at the speed of tool execution.
Read article