Apple isn’t making a chatbot… or are they?
Apple’s approach to AI seems different from what most of us expect. Rather than a standalone conversational assistant like ChatGPT or Claude, Apple’s AI is more of a toolbox: a collection of practical features integrated into its operating system, along with some creative AI tools. Accuracy is decent, usefulness is situational, and overall impact has been modest. But because these tools are baked into iOS, they quietly serve their purpose without fanfare.
Meanwhile, the AI that most users rely on today—ChatGPT, Google Gemini, Claude—are chatbots. Surprisingly, Apple has internally dismissed the idea, believing such bots offer “limited value” to users.
Yet behind the scenes, Apple has actually developed a chatbot. The catch? Most users may never see it.
Apple’s GPT: Siri’s Secret Training Buddy
Bloomberg reports that Apple’s internal chatbot project, codenamed Veritas—Latin for “truth”—exists solely for internal use.
Veritas looks like a standalone iPhone app, similar to ChatGPT: you can type questions in a chat interface, save and reference conversations, and have extended interactions.
What sets it apart is its capability to perform tasks within apps—editing photos, searching personal data, and more. In essence, it’s not just answering questions; it’s executing actions. This links directly to Apple’s delayed AI Siri project, with Veritas serving as a “training ground” for queries that don’t rely on voice input.
Back in 2023, Apple tested AppleGPT internally under the codename Ajax—a chatbot built on Apple’s own language model. Limited to internal use, it helped design product prototypes. While Ajax operated via web interface, Veritas is a standalone app—but Apple currently has no plans to release it publicly. Essentially, it’s Siri’s training partner.
Why Users Want AppleGPT
Naturally, users would love access to such a tool. But Apple sticks to its principles, often choosing caution over opportunity. Known for a conservative stance on generative AI, Apple has even resisted market pressures to embrace chatbots fully.
At this year’s WWDC, Senior VP Craig Federighi stated clearly that building a chatbot “has never been, and will never be” Apple’s goal.
Ajax was eventually shut down internally; access required special approval, and Apple forbade its outputs from being used for consumer-facing functions. Publicly, Apple has repeatedly downplayed the need for chatbots, arguing that users aren’t interested and that Apple’s AI position isn’t weakened by the lack of conversational bots.
Yet the global success of ChatGPT proves otherwise: chatbots are transforming how people communicate, learn, and create. Apple’s AI hasn’t matched this experience—so the Veritas leak reignites the question: does Apple need AppleGPT?
AppleGPT vs. Siri
Apple envisions Siri as the heart of its AI experience: speak, and an enhanced AI Siri performs commands contextually across apps.
Even if AI Siri launches flawlessly, it will likely lack the aggregated reasoning and conversational depth of a chatbot—features many users now expect from AI. Last year, Apple integrated ChatGPT into Siri, allowing questions to be routed through ChatGPT. But the experience is clunky: shallow integration, no continuous conversation, suitable only for simple queries. Users still end up opening ChatGPT directly.
Apple’s privacy-first approach and closed ecosystem make it hard for third-party chatbots to “take over” Siri. Still, Apple is quietly exploring internal solutions. Bloomberg recently reported a new internal team—Answers, Knowledge & Information (AKI)—working on an AI-powered “answer engine,” similar to Perplexity or ChatGPT’s search functions, planned for integration into Siri next spring.
This approach reduces risk: rather than a free-form chatbot, it aggregates existing web resources intelligently, limiting hallucinations while still providing useful answers.
Veritas may be more capable than this engine, akin to the difference between ChatGPT and Perplexity. Its leak has fueled calls for Apple to release it as a true AppleGPT—a chatbot integrated with Siri that meets user expectations and signals Apple’s commitment to AI.
But AppleGPT Isn’t Typical Apple
Launching a full-featured chatbot now could seem like following trends rather than innovating. Apple lacks a strong lead in conversational AI, and an underperforming AppleGPT could harm its already cautious AI reputation. Even top AI firms’ chatbots face hallucinations; Apple won’t release a product it can’t guarantee.
Instead, Apple’s strength lies in its platform. Debates continue internally about building vs. partnering. Using its ecosystem to set the rules for AI use aligns with Apple’s core strategy. Federighi compares it to the early internet: Apple didn’t need its own search engine or e-commerce site. The App Store itself succeeded because Apple provided the platform, not every service.
Similarly, AI on iPhones is about how users interact with models, not letting AI dictate Apple’s products. Apple is developing a universal AI interface (MCP) to connect devices with any AI model, enabling system-level operations from third-party AI.
At present, Apple’s priority isn’t AppleGPT. It’s deepening ChatGPT-Siri integration, adding more models, and letting Siri become a platform offering multiple AI responses—arguably more useful than a standalone chatbot.
