Sources: Apple's AI model team wanted to release several models as open source but Craig Federighi disagreed, largely concerned about public perception issues
Earlier this year, the Apple team working on the company's artificial intelligence models wanted to release several of them as open-source software.
Context & Ripple Effects
Apple's reported internal debate sits alongside earlier concerns that its own AI/ML organization might not deliver generative-AI capabilities, as described in internal doubts about Apple's AI team's readiness. The disagreement makes model distribution—not just model development—a leadership decision at Apple.
The reported stance also fits a more guarded AI posture, including restrictions on employees' use of outside generative-AI tools over confidentiality concerns. Later coverage of Federighi's push to use outside models for Siri suggests Apple was weighing control, product readiness, and partner dependence together.
First-order effects
- The AI model team's proposed open-source releases were reportedly halted, keeping those models under Apple's control rather than making them available to outside developers and researchers.
- Federighi's objection gives public perception a direct role in Apple's model-release decisions, alongside technical performance or product usefulness.
Second-order effects
- Without an Apple-released open model, developers cannot build directly on these reported models; Apple retains greater control over how its AI work reaches users and partners.
- The choice increases the practical importance of integration choices, including the later-reported push to use other companies' models for a revamped Siri, rather than relying on external adoption of Apple's own models.
Third-order effects
- If this pattern persists, Apple may compete in AI primarily through tightly governed product integration and selected model partnerships, not through an open-model ecosystem.
- The case illustrates a broader access-governance divide: companies may treat model release as a brand and risk-management decision as much as a developer-relations strategy.
The trend: AI companies are increasingly making model access a strategic governance choice, balancing ecosystem reach against reputational, safety, and product-control concerns.