Apple, OpenAI and Anthropic: What Powers Siri?

Yes—Apple has called on both OpenAI and Anthropic as potential providers of AI technology for a more capable Siri, according to reporting about its discussions with the companies. That does not mean both models are already powering Siri: Apple’s announced ChatGPT integration is a specific feature, while any Anthropic role has been reported as a proposal under consideration.

What Apple has announced

Apple introduced Apple Intelligence in 2024 as a collection of features for compatible iPhones, iPads and Macs, including writing tools, image generation and a more capable version of Siri. Apple said its own models would handle many requests, with ChatGPT available as an additional option when a request would benefit from broader knowledge or content generation.

The announced arrangement is not a wholesale replacement of Siri with ChatGPT. Siri remains Apple’s assistant and interface; Apple’s software decides when to answer itself and when to ask for permission to pass a request to ChatGPT. Apple said users would be asked before information was sent, and that ChatGPT access could be used without creating an account. Users who connect an account may receive the benefits and follow the terms of that service.

Apple’s public announcement named OpenAI and ChatGPT. Reports about Apple discussing models from Anthropic, including Claude, describe negotiations or options being explored—not a confirmed consumer feature. The distinction matters: discussions do not establish a signed deal, an integration date or a promise that a particular model will appear in Siri.

Why Apple may want more than one AI partner

Building useful assistants involves more than producing a fluent answer. Models differ in their strengths, operating costs, speed, reliability, privacy arrangements and ability to run on a device. Apple can use its own models for tasks suited to its ecosystem while evaluating outside systems for requests that need different capabilities.

Considering multiple providers can also give Apple leverage and flexibility. It may compare quality, performance and terms before deciding whether a model belongs in a product. A provider could be useful for some tasks but not others; the choice need not be all-or-nothing. Apple has not publicly laid out a final multi-provider plan for Siri, however, so this is context for the reported talks rather than a confirmed design.

• Capability: A large external model may handle some open-ended questions or writing tasks differently from a smaller on-device model.
• Privacy: Apple has emphasized on-device processing and its Private Cloud Compute approach for certain requests. Any external service raises separate questions about what data is shared and under what terms.
• Cost and speed: Running powerful models can require more computing resources, and responses may vary in latency and expense.
• Resilience: Having choices may reduce dependence on one supplier, though Apple has not said this is the reason behind any particular negotiation.

How the Siri and ChatGPT handoff works

Apple’s announced ChatGPT feature is designed as an optional handoff. A user can ask Siri a question, and if Apple’s system considers ChatGPT useful, Siri can offer to send the request. The user can also invoke ChatGPT directly in supported Apple Intelligence experiences, such as certain writing and image-related tasks. Availability depends on device, software, language and region.

When a request is handled by Apple’s on-device models, it can be processed locally where supported. More demanding requests may use Apple’s Private Cloud Compute, which Apple describes as a way to extend processing to server-based Apple silicon while applying privacy protections. A ChatGPT handoff is different: it involves OpenAI’s service, subject to the consent flow and applicable service terms. These paths should not be confused as one shared system.

Nor does an outside model automatically gain unrestricted access to a person’s device or private data. The intended experience is a controlled feature within Apple’s operating systems, with prompts and permissions. Still, users should read the on-screen confirmation and understand that sending a prompt to an external service means that service receives the information needed to respond.

What an Anthropic role would—and would not—mean

If Apple were to integrate Claude, that would make Anthropic another model provider available for selected tasks; it would not necessarily mean Claude replaced Apple’s models, ChatGPT or Siri itself. The exact interface, eligibility, privacy rules and supported features would depend on an announcement and product documentation. As of the reporting behind this question, those details should not be treated as established.

There is also a difference between a company considering a model and choosing it for a shipping product. Technology companies routinely assess suppliers and negotiate possible arrangements. Until Apple confirms an integration, claims that Claude is already built into Siri—or that Apple has finalized a deal with Anthropic—go beyond what the public announcement establishes.

What users should expect

For users, the practical picture is a layered assistant rather than a single chatbot taking over. Siri remains responsible for the Apple interface and device actions; Apple Intelligence supplies Apple-built capabilities; ChatGPT is an announced optional extension for certain requests. Other outside models could be added later, but no timetable or final arrangement should be assumed from reports alone.

Feature access is also not universal. Apple Intelligence requires supported hardware and software, and Apple rolls out features by language and region. A user who does not see a feature should check Apple’s current compatibility and availability information for their device and location rather than assuming the feature is missing because of an account problem.

To make a sensible privacy choice, consider what a prompt contains before approving a handoff. Avoid sending confidential work, personal identifiers or other sensitive details unless you understand the service’s applicable privacy terms and have permission to share them. The permission prompt is meaningful: it marks a change from a request handled within Apple’s system to one sent to an outside provider.

Frequently Asked Questions

Is ChatGPT already part of Siri?
Apple announced an optional ChatGPT integration for Siri and other Apple Intelligence features. It is a handoff for suitable requests, not a replacement of Siri by ChatGPT, and availability depends on supported devices, software and regions.

Has Apple officially chosen Anthropic for Siri?
Apple publicly announced ChatGPT from OpenAI. Reports of talks with Anthropic do not, by themselves, confirm a finalized deal or a Claude feature in Siri.

Will Siri use OpenAI and Anthropic at the same time?
That is possible in principle, but Apple has not publicly confirmed a final system in which Siri routes requests between both providers. The announced external integration is with ChatGPT.

Does Siri send every question to ChatGPT?
No. Apple describes ChatGPT as an optional route for requests that may benefit from it. Users are asked before a request is shared through the integration.

Can I use the ChatGPT integration without an account?
Apple said users could access the integration without signing in to a ChatGPT account. Signing in is optional and may bring the account’s own settings and service terms into play.

Does Apple Intelligence work on every iPhone?
No. Apple Intelligence is limited to supported hardware and software, and feature availability can vary by language and region. Check Apple’s current compatibility details for the specific feature you want.

Related

• Pokémon GO Spotlight Hour in February: Dates and Details
• No Backend Meaning in the Dev Community Story
• Galaxy S26 vs Z Fold8 Ultra: What $999 Buys