Apple is reportedly in talks with AI companies Anthropic and OpenAI to power a revamped Siri, signalling a major shift in its approach to artificial intelligence and a rare admission that its in-house tech may be falling behind.

According to sources close to Bloomberg, Apple is testing custom versions of Claude by Anthropic and ChatGPT by OpenAI on its own Private Cloud Compute infrastructure.

These trials come as Apple grapples with delays and internal discord surrounding its Apple Foundation Models – the engine behind its current AI tools.

If Apple proceeds, it would mark a major reversal for the company which has historically prioritised in-house innovation.

The move would also mirror rivals like Samsung, which uses Google’s Gemini model under the Galaxy AI brand, and Amazon, which has integrated Anthropic into Alexa+.

Apple intelligence Siri

The shift comes after Siri’s leadership was handed to Mike Rockwell, who is best known for launching Apple’s Vision Pro, and software chief Craig Federighi. They replaced AI head John Giannandrea, whose oversight has been scaled back after its latest AI efforts failed to make an impact.

Despite launching new Siri features last year, including contextual app control and on-screen content analysis, the rollout has been repeatedly delayed. The next major update is now expected in 2026.

Internally, Apple has acknowledged its models may not be competitive with leading offerings from OpenAI or Anthropic. Executives are reportedly warming to the idea of licensing external tech as a near-term fix, with Anthropic emerging as a favourite. Pricing remains a sticking point though, with Anthropic reportedly seeking a multibillion-dollar annual deal.

While Apple is committed to maintaining control over on-device AI models, which underpin features like Genmojis and email summaries, its openness to outsourcing key Siri capabilities marks a notable strategic shift.