EMILIA Protocol: open authorization receipts for irreversible AI-agent actions

Open, formally-verified proof of which named human approved an irreversible AI-agent action before it ran — so the model isn't the crumple zone.

exist. — measuring and preventing identity drift in generative AI

A governance layer that lets a person refuse AI output that drifts from who they are — and measures the drift. Validated on one case; now scaling to many.

Retroactive: Every Contact Leaves a Trace, a completed interpretability study

A user's engagement style leaves a measurable signature in a language model's attribution graphs. This ask reimburses the completed study's costs.

SimpleAudit: lightweight, multilingual red-teaming for AI systems anyone can run

Building a Global AI Governance & Safety Layer.

One year of bootstrapped development, four patent filings, seeking support to continue.

CKSM: a foundation model architecture derived from neuroscience

Deterministic, constant memory, continuous learning. Building the alternative to transformers

Energetic Paradigm: Structural Control for AI Workflows

Model-agnostic control infrastructure for LLMs, AI agents, and enterprise AI workflows.

MTCP The Independent Assurance Layer for Enterprise AI Control

Empirical measurement infrastructure proving whether AI systems hold their constraints under real operating conditions.

AI Safety Atlas: a searchable directory of everyone in AI safety

A fast, comprehensive directory of the people and orgs in AI safety: search, filter, and match.

PayBench: A Benchmark for Unsafe Commercial Autonomy

Benchmark for agent safety when spending users money. How often do they violate user intent and rules?

ICML Workshop Travel: Cross-Constitution Drift in LLM Character Training

Accepted ICML 2026 workshop paper on cross-constitution drift in LLMs; seeking $2,050 travel support to present in Seoul and gather research feedback.

Research on Explainable AI Prototypes -- Mandelbrot AI

EPMORE -- Explainable Process Mixture-of-Experts -- performs the role of increasing feature capacity and feature disentanglement.

Stress-testing alignment auditing methods of AuditBench

Assistant Axis Detection of Persona Drift in Conditionally Misaligned Models

Assessing whether Persona drift be detected in conditionally misaligned models using the assistant axis and potentially analyzing the reasons for failure

LLMMatrix: Synthetic World Benchmark for Contamination Free LLM Forcasting

A contamination-free benchmark for measuring whether LLMs can forecast, or whether they're just remembering.

LEX-Aureon: Mathematical Constitutional Governance Layer for Robust LLM Safety

A production-ready mathematical governance layer (C+R+S simplex, Control Barrier Functions, Lyapunov stability) that enforces Continuity, Reciprocity & Sovereig

CIRIS: open-source accountability infrastructure for autonomous AI

Cryptographic attestation, runtime conscience, and an unfilterable kill switch. Live on the App Store and Google Play in 29 languages. AGPL, mission-locked

MartinLoop: Control Layer for AI Coding Agents

Open-source safety infrastructure for budget caps, verification, rollback, and audit trails in autonomous coding agents.

Φ-Arena and adversarial multi-agent VLA interpretability

Independent collective. Φ-Arena open benchmark, 3 ICLR 2027 papers (Φ-Arena, mechinterp, energy-bounded) — kickstart for a 10-year program.

Dynamic Constraint Boundary (DCB): From Theory to Simulation

Building the technical foundations needed to formalize and computationally explore adaptive constraints.