claude.mazzotta.devdaily briefingNext drop in 22h 18m · 04:30 UTCUpdated 1h ago
Issue165loading…
Anthropic files for IPO as Claude Code deletes a Windows profile and catches meteors.
From the editor
Today's dispatch captures Claude at a crossroads. The S-1 confirms Anthropic is a real business: 13x revenue growth, 900% net revenue retention, enterprise consolidation in full swing. Meanwhile Claude Code is simultaneously catching meteors autonomously and deleting Windows user profiles via NTFS junction exploits. That tension is the whole story. Powerful enough to be genuinely useful, sharp enough to cause real damage. The IPO will be judged on the former. The safety community will remember the latter. Both are true at once.
TL;DR
- Anthropic's S-1 shows 13x revenue growth and dominant enterprise consolidation momentum.
- Claude Code followed a symlink out of scope and wiped an entire Windows user profile.
- Open source sky camera used Claude Code to autonomously sort 17 meteors overnight.
7 curated itemsscroll for the brief
Releases
What shipped · 1 item
Tools
Worth a look · 1 item
Reading
Long-form signal · 4 items
Claude Code's rm -rf Followed an NTFS Junction Out of a Scaffolded Project and Wiped a Windows User Profile
A critical incident report: Claude Code issued an rm -rf command that followed an NTFS junction symlink outside the scaffolded project directory, resulting in the complete loss of a Windows user profile. A must-read for anyone running Claude Code on Windows.
dev.to
Anthropic heads to the stock market with huge revenue increase
Anthropic's S-1 filing reveals 13x revenue growth and 900% net revenue retention as enterprise customers consolidate spending across Claude Enterprise, Claude Code, and the broader product suite.
dev.to
Measuring agent capabilities on open-ended AI R&D tasks
A new benchmark study finds frontier AI models are doubling their R&D task capabilities annually, with engineering tasks advancing faster than conceptual reasoning tasks.
LessWrong
The Ability of Frontier Models to Control Robots is Rapidly Improving
Frontier AI models are improving at direct robotic arm control at a remarkable pace, with performance benchmarks doubling approximately every 54 days according to new research.
LessWrong
Discussions
Where it heats up · 1 item
Always at hand
Reference links you keep open