Siri’s Next Chapter Could Be Powered by OpenAI
Toggle Dark Mode
Apple may abandon its plans to use its own in-house AI models to power the next generation of Siri, opting instead for third-party partnerships to enhance its voice assistant.
It’s no secret that Apple has been struggling to deliver on its promised improvements to Siri. The company got everyone’s hopes up at last year’s Worldwide Developers Conference (WWDC) when it demonstrated a Siri that would be more capable and better at understanding personal context. Such a development would have been a dream come true for many iPhone users, allowing the digital voice assistant to function with the efficiency of a human assistant by using personal context to respond to simple requests.
Apple gave an example of this during its WWDC 2024 keynote by asking, “What time does my mom’s flight land?” From there, Siri would ferret out the information on who the user’s mother is, check emails and messages for recent flight information, and then use a flight tracking site to determine the estimated time of arrival.
Although Apple almost certainly made these promises with the highest of hopes on its ability to deliver, it failed to do so. Earlier this year, it became apparent that what Apple had shown at WWDC 2024 didn’t exist in a demonstrable form, and the company officially announced that the new Siri wouldn’t make the cut for iOS 18 after all. Now, it’s not expected to arrive until early 2026.
Apple’s software chief, Craig Federighi, was reportedly given command of the Siri project to get it back on track after Apple’s AI and machine learning department failed to deliver on time. Following Apple’s 2025 WWDC keynote earlier this month, Federighi explained that they had to move to a new “V2” architecture as the limitations of its original architecture weren’t delivering the required quality of responses.
Earlier reports suggested that the new Siri was entirely wrong about a third of the time (although some might call that an improvement), and while Federighi didn’t elaborate on those, he said that what Apple had would not meet customer expectations or Apple’s standards. Apple’s executives confirmed that the Siri improvements wouldn’t arrive before 2026.
Such a move has left Apple falling even farther behind its competitors, like Gemini Live and ChatGPT. While Apple can probably crack this nut eventually, time is a luxury the company doesn’t have. If a personal Siri can’t be perfected until 2026, then a more conversational Siri will likely take until at least early 2027. Google introduced its more conversational version of Gemini Live last year, so imagine where it will be by the time Siri is ready to have a normal chat with you?
So, it’s no surprise that Apple is at least exploring the possibility of bringing in some outside help. While there’s no indication that Apple has thrown in the towel just yet, Bloomberg’s Mark Gurman reports that Apple is talking about using OpenAI or Anthropic to power Siri 2.0.
These discussions go beyond internal chats among Apple’s senior leadership. Last week, Gurman reported that Apple was considering a multi-billion-dollar bid for Perplexity AI, but those discussions hadn’t moved beyond Apple’s boardrooms. Apple and Perplexity have discussed a potential partnership. Still, sources say this may be more about search than Siri, partly as a way for Apple to protect itself if its lucrative Google search deal is nixed by the US Justice Department’s antitrust case against Google.
In this case, insiders say that Apple has spoken with both OpenAI and Anthropic about using their large language models specifically for Siri. It’s even gone so far as to ask them to work on models that could run on Apple’s Private Cloud Compute infrastructure.
Such a move would be a significant shift in Apple’s strategy and a tacit acknowledgment that it can’t compete in generative AI. However, there’s reason to believe Apple may simply be exploring possibilities here, setting a “plan B” if you will. These investigations are still at a very early stage, and the internal LLM Siri project is still moving full steam ahead. However, with the struggles Apple has encountered thus far, it would be foolish for the company not to have a backup plan. Apple cannot afford to fall any further behind, and if LLM Siri V2 doesn’t produce acceptable results, it’s unlikely that there will be a V3.
[The information provided in this article has NOT been confirmed by Apple and may be speculation. Provided details may not be factual. Take all rumors, tech or otherwise, with a grain of salt.]
