News

Latest from DiffLab

Research updates, product launches, and insights from our team.

September 11, 2026strategyai-transformationguide

Announcing the AI Strategy Guide

DiffLab publishes the AI Strategy Guide — a practical plan for doubling the productivity of the people at your company, sequencing the change, and sharing the result.

September 10, 2026researchmodelingopen-source

10B Model: People not Parameters

An open-source model of high-level human needs — part of our approach to modeling and simulation, so agents can help answer what problems a business or community can tackle next.

April 15, 2026talkai-transformationsoftware-factories

Talk: AI Software Factories

A talk on AI software factories — how organizations industrialize AI-native software development with custom control surfaces and production systems.

March 19, 2026toolsagent-securityopen-source

prompt-bonk: A Sub-Millisecond First Layer Against Prompt Injection

If you're building agentic systems that process untrusted input, you need a fast first filter before anything reaches your LLM. prompt-bonk is a zero-dependency regex scanner that runs in under 1ms and catches the attacks that break your system — not the ones that produce rude text.

March 11, 2026researchboundary-enforcementpublication

The Rule Problem: Key Hazards in AI Boundary Enforcement

Safety, alignment, and boundary enforcement are as much a system problem as a model problem. Our new research preview maps four ways state of the art AI boundary enforcement breaks down — and what to do about it.

March 6, 2026researchcognitive-securitytuneout

Introducing TuneOut: A Practical Guide to Cognitive Security

DiffLab publishes TuneOut, an open research guide to cognitive security — practical techniques for reclaiming your attention from algorithmic manipulation using AI-powered information filtering.