Phenogenesis: Preserving Human Judgment in the Age of AI

A six-month pilot testing how AI can strengthen human reasoning without displacing judgment, responsibility, or ownership of decisions.

KernelBench: an independent benchmark for AI-automated GPU kernel engineering

Frontier-model sweeps, hardware-roofline scoring, and reward-hacking audits at kernelbench.com. Keeping an independent AI R&D automation benchmark alive.

IRL-2: Provably Convergent Iterative Reasoning in Sparse Transformers

A fixed-point reformulation of iterative reasoning in sparse MoE models, replacing a fixed step count with provable convergence. Builds on a validated v1

Rewind: Atomic Rollback Infrastructure for AI Agents

Ctrl+Z for AI agents. Rewind automatically snapshots the filesystem and the conversation together, so when your agent breaks something, one rollback fixes both.

ValiChord: verification for AI evals and science that can't be gamed

Independent validators test a claim blind, seal their verdicts, and reveal together — a permanent, tamper-proof record of AI evals and scientific results.

Noisify: adversarial photo protection from AI nudification

A web service that makes your photos resistant to AI-powered non-consensual undressing

An Open Authority Standard for Embodied AI Agents That Act on the Physical World

An open schema, consent framework, and harm taxonomy bounding what embodied AI agents may do in the home, validated on a shipping humanoid robot.

What breaks when a robot never stops learning? Continual RL on real hardware

A Dreamer 4-style multimodal world model with a grounded, self-generated reward, learning 24/7 on a $500 open arm — measuring wireheading, forgetting, drift

Emotion as Attack Surface and Monitoring Signal in Thinking Models

ClauseHound

Indus AI Node

Pax Machina

A publication about the institutions we need for powerful AI.

Making sure that the code AI pushes to the internet is safe and bug-free

Building LemmaScript, a verification toolchain for TypeScript

Finding and Defending the Attention Circuits That Make LLMs Jailbreakable

Mapping the attention heads that push LLMs toward refusal vs. compliance, and building an inference-time defense against both single- and multi-turn jailbreaks.

Travel grant to attend HAAISS Prague 2026

BeOriginal.id Before You Click. Before You Scan. Before They Steal.

Open infrastructure helping anyone detect malicious links, fake QR codes and brand impersonation before becoming a victim.

Honest evals and provenance for AI-agent memory

An open test of whether AI-agent memory systems spread false claims, plus a provenance spec so an agent's memory can be audited, corrected, and forgotten.

Individuation without Objectives — measuring how identity forms without goals

Physics verification for AI-generated hardware designs

A model already designed a 44-piece engine inside the verify loop. This funds measuring how reliably it does that, and scaling to harder machines.

Beyond Asymmetric Control: Bypassing Systemic Friction in AI Alignment via

Ontoresonant Feedback Loops

Imbas: the inspection layer for AI answers

A live Reader that inspects what AI answers surfaced, missed, and shaped, turning each run into a record for a future inspection agent.