According to Bloomberg’s Mark Gurman, engineers inside Job’s Mob are quietly panicking over the performance of the AI-enhanced Siri set to arrive with iOS 26.4 in spring 2026.

While no specific failure points were cited, past leaks suggest Siri still fumbles basic tasks when asked to operate within apps or deal with critical use cases like banking.

Apple has been pushing flashy features like in-app actions and personal context awareness, but whether Siri can actually handle them without choking is another matter entirely.

This comes after years of excuses blaming Siri’s original framework for its sluggish development. Craig Federighi had previously blamed everything on Siri’s “V1 architecture” and promised a glorious V2 overhaul. That second version is now reportedly live behind the scenes and still floundering.

Among the promised upgrades is the ability to dig through your personal data for smarter responses. Siri, in theory, should be able to find a podcast mentioned in a text thread or add items to a list via voice. So far, though, it appears the assistant is just as clueless with context as ever.

In what looks suspiciously like a vote of no confidence, Apple AI exec Ke Yang, who was leading the Answers, Knowledge and Information (AKI) team tasked with bringing web search to Siri, has already bailed for Meta. Yang was only appointed weeks ago but apparently saw enough to make a quick exit.

The AKI team was meant to give Siri something vaguely ChatGPT-like, with the ability to pull useful info from the web. With Yang gone and Gurman’s latest scoop pointing to continued dysfunction, it’s fair to ask whether Siri V2 is just another shiny wrapper around the same old mess.

If this is what Apple’s AI ambitions look like in practice, then Job’s Mob might want to stop mocking everyone else’s chatbots and get its own house in order.