Protecting Intellectual Priority in the Age of AI: Runnable Encrypted Code and Public Provenance for Private Repositories

The rise of AI-powered coding assistants introduces two new threats to researchers and developers working in private repositories: inadvertent content leakage through AI tools that read local files, and difficulty proving independent development when concurrent work appears publicly. This working paper presents two complementary tools. Shadow Archive provides timestamp evidence for private Git repositories by maintaining an obfuscated hash tree in a separate public repository. AiLock offers in-place file encryption with memory-only execution: files remain ciphertext in the working tree, while encrypted Python code is decrypted only inside a controlled runtime so it can still be run, imported, and inspected. Together, the tools form a lightweight toolkit for protecting intellectual priority in AI-assisted software development.