Story
NeurIPS is tightening the social contract of AI research
Reproducibility, disclosure, and responsible research are moving closer to the center of conference operations.
Topic dossier · Reviewed
AI Safety
Evaluation, alignment, security, and governance of advanced AI systems.
Editorial scope
AI Safety
Evaluation, alignment, security, and governance of advanced AI systems.
Essential coverage
Story
Thinking Machines Lab is organizing around collaboration
The new lab’s public thesis places human control and customization at the center of advanced AI.
Story
Safe Superintelligence is narrowing the company to one problem
A singular mission can clarify priorities—and postpone the questions that deployment usually forces.
Story
The next developer stack includes evaluation by default
Prompting is giving way to traces, datasets, tests, permissions, and rollback.
Feature
The people who decide whether an agent is ready
Evaluation teams are becoming the quiet editorial layer inside autonomous software.
Feature
The company page as an accountability surface
In the AI economy, products change faster than institutions can explain themselves.
Explainer
What does an AI agent actually do?
A practical guide to planning, tools, memory, permissions, and human review.
Explainer
How are AI systems evaluated before release?
Benchmarks are only one layer in a release process that also needs scenarios, experts, and monitoring.
Explainer
Why do AI models hallucinate?
Fluent generation is not the same operation as checking a claim against the world.
Explainer
How does retrieval-augmented generation work?
RAG gives a model selected documents at answer time, but it does not guarantee that the answer is correct.
Explainer
What changes when software can use tools?
The moment a model can send, buy, edit, or deploy, interface design becomes authorization design.
Research
A field guide to frontier model documentation
We compared the public artifacts used to explain capability, safety, and release decisions.
Research
How AI companies describe safety
A taxonomy of public commitments across labs, infrastructure providers, and application companies.
Research
An index of AI conference disclosure rules
NeurIPS and ICML are turning tool use, reproducibility, and research ethics into explicit publication operations.
Podcast · Signal Path
Signal Path 06: Voice becomes an interface
Synthetic speech is moving from media production into agents and accessibility.
Film
The room where agents are reviewed
A transcript-first visual essay about traces, tests, permissions, and the decision to ship.
Profile
Dario Amodei and the institution built around scaling
Anthropic’s CEO has made capability, risk, and organizational design parts of the same public argument.