Cal.com announced it is closing its previously open-source codebase after detecting large-scale AI scraping intended for training proprietary models. Primary reporting from Strix.ai (2024) states the company viewed the extraction as an existential threat that allowed competitors to replicate features without upstream contributions. This aligns with documented cases including the 2022 class-action suit against GitHub Copilot for using open repositories without attribution (Northern District of California) and HashiCorp's 2023 adoption of the Business Source License reported by The Register to counter cloud-provider exploitation.

Original Strix coverage frames the move as Cal.com learning the "wrong lesson," yet omits the structural pattern of enclosure: permissive MIT/Apache licenses have enabled foundation-model trainers such as OpenAI and Anthropic to ingest communal code at scale while releasing closed weights, a dynamic analyzed in Stanford HAI's 2023 report on AI data governance. The HN thread (item?id=47780712) further reveals community recognition that technical countermeasures like dataset watermarking, discussed in a 2023 NeurIPS paper on adversarial dataset protection, remain immature and unenforceable at repository scale.

Synthesizing these sources shows an accelerating shift from open collaboration to defensive closure across infrastructure projects. Without new licensing norms or collective rights-management mechanisms, the uneasy relationship between shared code repositories and proprietary model training risks shrinking the open commons that AI itself depends upon.