Source Code Leak
Stories from Source Code Leak come first, newest first, then stories from connected topics (Claude Code, Anthropic, AI), each labelled with where it comes from. This view is not indexed by search engines.
Explore related topics in the Atlas →
Anthropic Tried to Kill 8100 GitHub Repos. Then This Happened.
Anthropic DMCA'd 8100 GitHub repos trying to contain the Claude Code leak. A developer rewrote it in Python before sunrise. 110K stars. The fastest repo in GitHub history. The genie is out of the bottle.
Latest
US Navy Says a Carrier Strike Group Has Deployed With an Unmanned Surface Vessel
Acting US Navy Secretary Hung Cao says the Theodore Roosevelt carrier group deployed on 8 October with an unmanned surface vessel, a first. Model unnamed.
Epoch AI Says Open-Weight Models Trail Closed Ones on Cyber Skills by 6-7 Months
Epoch AI estimates open-weight models trail closed ones on cyber skills by 6-7 months, with Z.ai's GLM-5.3 on top, but its score rests on three results.
Goodfire Essay Claims Classic Superposition Evidence Was Misread
Goodfire’s Nick Cammarata argues neural nets are already mostly clean and classic superposition evidence was misread. Informal claim; critics push back.
Tesla Driver Tells CHP He Was Asleep in Driver-Assist Mode When He Hit a CHP Car
A Tesla hit a CHP patrol car at an I-80 crash scene near Fairfield, hurting two officers. The driver says he was asleep in driver-assist mode; unconfirmed.
Anthropic Says Claude Designed More Efficient Quantum Circuits
Anthropic says Claude Mythos 5 found more efficient quantum circuits than published designs. Self-reported research claims; not independently replicated.
Microsoft and NVIDIA Publish a Risk Guide for Open-Weight Models in Enterprise
Microsoft and NVIDIA's joint paper on open-weight model risk sets out five risk categories and argues openness alone does not decide risk. More to come.
NIST, NSA and CISA Issue 10 Voluntary Practices for Super Intelligence Incidents
NIST CAISSI, NSA's SISC and CISA set out ten voluntary practices for incidents where AI agents act outside instructions in training and testing.
Intuit Signs Deal to Acquire AI Revenue-Management Startup Sequence
Intuit signed an agreement to acquire Sequence, an AI-native revenue platform. Terms undisclosed; ~$200m reported by The Times via Tech.eu. Not closed.
Bristol Lab-Grown Brain Organoids Decode Braille Tactile Signals
Bristol researchers showed brain organoids can pass Braille tactile signals a decoder can read. Peer-reviewed; they did not demonstrate learning.
Elorian Releases Glimpse, Its First Visual-Reasoning Model
Elorian opened early access to Glimpse, its first visual-reasoning model. Details via TechCrunch; benchmark wins are Elorian’s own evaluations.
Latest on TSN
Explore the TSN Atlas: Every topic we cover, mapped. Click a bubble to dive in.