Human-AI IntersectionSimon Willison

tldraw issue: Move tests to closed source repo

Read original

A comprehensive test suite is enough to build a completely fresh implementation of any open source library from scratch, potentially in a different language

Key takeaways

  • AI coding tools can use comprehensive test suites to reverse-engineer entire codebases, creating new IP vulnerability for open source projects
  • tldraw is moving tests to private repos as defensive measure, signaling broader concern about AI-enabled code replication
  • Open source business models built on 'open code + commercial support' may need fundamental rethinking as AI agents can replicate implementations from test specifications alone
  • The joke issue about translating code to Traditional Chinese highlights absurdity of trying to defend against AI through obfuscation
  • This represents a new category of AI impact: not replacing developers, but undermining open source sustainability models

I cover AI×GTM intelligence like this every Wednesday.

Get STEEPWORKS Weekly

More picks

GTM OpsSales and SellingVictor's pick

6 Months Job Hunting. Finally Landed an AE Role

massive anecdonetal story of the shift structurally in GTM, we should pay attention to this one

  • Sales job market in 2025 showing severe compression: 80 interviews across 36 applications over 6 months, with only 4 reaching final rounds (11% final-round conversion rate)
  • Salary deflation evident: multiple 'reputable companies' offering <50% of previous AE compensation, forcing candidates to reject offers that pay less than gig work
  • Entire sales teams being managed out simultaneously (August 2025), suggesting widespread GTM restructuring or company failures, not individual performance issues

This analysis was produced using the STEEPWORKS system — the same agents, skills, and knowledge architecture available in the GrowthOS package.