Anthropic Open-Sources J-Lens Interpretability Tooling with a Neuronpedia Demo

Anthropic (release notes)·low signal

As part of its early-July updates, Anthropic open-sourced J-lens — interpretability tooling for inspecting model internals — paired with a live Neuronpedia demo, alongside restoring Fable 5 and Mythos 5 globally after US export controls were lifted. Useful for researchers and builders doing mechanistic-interpretability or model-debugging work who want hands-on tooling rather than papers. Included as a low-importance/older-than-48h item since it predates this run's core window but wasn't previously covered.