Apple is quietly running Veritas, an internal chatbot used by employees to test how Siri might evolve. This isn’t a consumer tool. Apple seems to be using it as a controlled sandbox before rolling out AI upgrades to the public.
What Veritas Does (and Doesn’t Do)
According to Bloomberg’s Mark Gurman, Veritas lets Apple employees interact as they would with ChatGPT or Google Gemini: you can send queries, have back-and-forth dialog, revisit previous threads. It tests features like:
- Searching personal data on the device
- Performing in-app actions (e.g. photo edits)
- Handling conversational context across multiple prompts
These experiments feed into the redesign of Siri, especially its “proactive assistant” ambitions.
Importantly, Apple has no current plan to release Veritas beyond its internal team. Gurman argues that’s a missed opportunity for real-world feedback.
Why Apple Is Keeping It In-House
Apple’s approach shows caution. Rolling out a public chatbot prematurely risks exposing flaws, weak privacy safeguards, or user dissatisfaction. The Veritas phase lets Apple test deeply within its own ecosystem before exposing it externally.
But critics see this as overcautious. Gurman suggests that by refusing public access, Apple might delay learning how real users interact with the system losing valuable signals in the process.
Meanwhile, Apple may lean on Google’s Gemini to fill gaps in AI search. Reports suggest Apple is considering a “hybrid” approach, combining its own models with external ones like Gemini or Claude.
Challenges and Timeline
Apple’s AI push has already faced delays. The enhanced Siri and Apple Intelligence features have seen lukewarm reception and repeated setbacks. Veritas may accelerate internal insight, but the real test lies in moving from experiment to polish.
If Apple waits too long, it risks falling further behind in the AI arms race. At this point, analysts expect the public version of next-gen Siri (or its AI-enhanced future) to arrive around iOS 26.4 in early 2026.
