Apple engineers are now allowed to leverage third-party large language models (LLMs) to enhance Siri's capabilities.
Craig Federighi is leading the initiative and has instructed Siri's machine-learning engineers to use open-source models from other companies if it helps in building better AI features.
Previously, engineers could only use third-party LLMs for testing, not in shipping features.
The move is aimed at improving Siri's performance as Apple's own models were found to be inferior to OpenAI technology.