ai-assuranceSep 21, 2026
Everyone Has an AI Policy. Almost Nobody Has Evidence.
AI assurance is becoming the pentest of the AI era. What it is, why it matters for anyone shipping AI into production, and what an honest one looks like.
Insights from the Lab
Our team thinking out loud about what happens when AI features meet real users. Incident analysis, field notes from assurance work, original research, and the occasional argument with the industry.
Incident Analysis
When an AI system fails in public, what actually went wrong underneath the headline — and whether yours has the same shape.
Assurance Practice
Field notes from auditing live LLM features. Adversarial testing, evals that survive a model change, and what buyers have started asking for.
Research & Benchmarks
Original data from our own studies and open-source benchmarks. Interpretability, evaluation, and the psychological safety of conversational AI.
Thinking Out Loud
Reflections from the ground between clinical practice, AI engineering, and running a small lab in Liverpool.
keido-labsJul 6, 2026
The bench and the judge. Local stack to keep AI psychologically safe.
Open-source pscyhology safety toolset for conversational AI
keido-labsJun 9, 2026
AIPsyBench - the way to keep AI conversations safe
open-source pscyhology safety toolset for conversational AI
ai-psychologyMay 9, 2026
What happens when you strip a language model of its feelings
A 9D Lesion Study in Emotional Subspace in Transformers
ai-psychologyMar 25, 2026
Whether, Not Which. What happens inside LLM when it process emotions.
A Mechanistic Dissociation of Affect Reception and Emotion Categorization in LLMs
keido-labsMar 8, 2026
Empathyc 1.3 Released
Empathyc 1.3 is live. And it's obsessed with safety and privacy.
Subscribe to Newsletter
Clinical psychology for AI. Research, insights, and frameworks for building emotionally intelligent systems.