(Zain bin Awais/PCMag Composite;Apple;Anthropic)
Apple reportedly is in early talks with Anthropic (Claude) and OpenAI (ChatGPT) to use one of their large language models (LLMs) to supercharge Siri. If this is indeed true, it would mark a major shift in Apple’s approach to AI: opting for third-party model integration instead of in-house development. Sure, you can access ChatGPT through Siri right now, but building Siri from the ground up with ChatGPT tech is a totally different story.
I don’t think Apple should choose ChatGPT to power Siri. Rather, it should go with Claude because of Anthropic and Apple's shared focus on user privacy. Both potential partnerships face plenty of headwinds, of course, but here's what you need to know.
Why Does Apple Need Another Company’s AI?
Let's back up. Why does Apple need to partner with an AI company in the first place? After all, Apple is famous for its walled garden approach to both hardware and software, stubbornly relying on in-house tech even when third-party solutions could be better. Beyond that, Apple already has a massive suite of AI features via Apple Intelligence.
The problem comes down to chatbot functionality. Apple has successfully leveraged its LLMs to enable features like Live Translation and Writing Tools, but it hasn't managed to follow through on its promise of a next-gen Siri. That version has faced many delays since it was revealed in June 2024.
Apple’s SVP of worldwide marketing, Greg Joswiak, has spoken about the delay. “We don’t want to disappoint customers," he said. "We never do, but it would have been disappointing to ship something that didn’t hit our quality standard that had an error rate that we felt was unacceptable."
Creating competent, mature chatbot models that can compete with the likes of Claude, ChatGPT, and Gemini requires many iterations and presumably a number of years. This simply won’t suffice for a feature that customers expect to have now. It seems unlikely that Apple can make up the necessary ground anytime soon. So, if Apple wants to ship an AI-powered Siri that doesn’t disappoint, it needs help.
Claude Is the Better Ideological Fit
Neither Apple nor Anthropic is perfect. Both companies have dealt with scandals and made decisions that people don’t like. However, both companies consistently dedicate themselves to protecting their users’ privacy. For example, Apple famously refuses to unlock iPhones for the government and employs the Private Cloud Compute for Apple Intelligence, while Claude doesn't train its models with user data by default.
This philosophical alignment makes these companies a great match. Losing the current ChatGPT integration in Siri wouldn't be a big deal if Apple and Anthropic could deliver an experience that's both capable and privacy-respecting.
(Credit: Anthropic/PCMag)Whereas DeepSeek is all too happy to funnel user data to foreign governments (according to a US House Select Committee report), and Gemini continues to introduce increasingly invasive features, a privacy-conscious chatbot on a privacy-conscious platform would be a major win for consumers. Although ChatGPT isn’t quite as bad as DeepSeek or Google, it doesn’t have a great track record when it comes to privacy. Someone hacked OpenAI in 2023, which the company didn’t even report until the next year. Chinese hackers continue to target OpenAI, too.
Beyond its vulnerabilities, OpenAI collects lots of user data and uses it to train its models by default. OpenAI is a massive corporation with ambitions far beyond selling chatbot subscriptions, too; it wants to partner with Apple alum Jony Ive to create AI gadgets. Why should Apple help what could soon be a direct competitor further refine its underlying technologies with Apple user data? Anthropic, on the other hand, doesn’t pose the same risks.
It Still Might All Come Down to Money
Even though an Apple and Anthropic partnership would be great news for online privacy, that alone might not be enough to make it happen. Besides, Apple already allows users to query ChatGPT through Siri, as mentioned. Since Apple has a preexisting relationship with OpenAI, expanding ChatGPT’s functionality within Siri could very well be the path of least resistance.
Relationships aside, businesses are about making money, and a ChatGPT partnership is more attractive financially. As a bigger company that operates its chatbots at a larger scale, OpenAI charges less for API access than Claude: $2 (input) and $8 (output) per million tokens versus $3 and $15, respectively, for Claude. So, OpenAI might be able to offer Apple a better deal. That said, Anthropic could always cut its rates, too.
We'll have to wait for Apple’s ultimate decision, but media accounts of the company's thinking are encouraging. Bloomberg reports that Apple executives consider Claude’s LLM performance the most promising when it comes to working with Apple tech. But that doesn’t mean ChatGPT and Claude are the only options. Apple reportedly has also discussed acquiring Perplexity. For now though, all there is to do is hope that companies that care about online privacy stick together.