The news: Apple’s highly anticipated AI enhancements, particularly for Siri, remain unfinished. During WWDC 2025, SVP Craig Federighi confirmed delays, stating Apple needs “more time to reach a high-quality bar.” No major voice assistant upgrades were announced.

Apple’s most relevant AI move wasn’t a product—it was a warning: Ahead of its event, Apple published a research paper arguing that top models like OpenAI’s GPT-4o, Anthropic’s Claude 3.7, and Google’s Gemini don’t truly “reason.” Instead, they create an “illusion of thinking.”

• The paper stated that reasoning AI models “fail to develop generalizable problem-solving capabilities, with accuracy ultimately collapsing to zero beyond certain complexities across different environments.” 
• The paper’s timely release could be taken as a preemptive shot to temper the industry’s focus and investments on AI.

Developers gain on-device AI access: Opening its on-device large language model (LLM) means proprietary features like mail summaries can be brought to third-party email apps, not just Apple Mail.

Early adopters include Kahoot (study guides) and AllTrails (personalized hike suggestions). Automattic’s Day One app will also integrate privacy-centric AI features.

Other AI takeaways: 

• Siri: Apple promised AI voice assistant updates “in the coming year.”
• Workout Buddy: The watchOS 26 app combines users’ activity logs, an AI chatbot, and real-time insights. This was one of few native AI updates.
• Visual intelligence: Apple is expanding this feature to allow users to run searches on any item appearing on their screens. Users can query OpenAI’s ChatGPT or run visual search on Google or Etsy.

ChatGPT remains the only generative AI (genAI) chatbot integrated into Apple’s ecosystem, and beyond visual search, it’s being used for online queries and translations. 

Our take: Apple is hedging its AI bets by being cautious with core offerings like Siri while quietly enabling developers with on-device LLMs and privacy-first tools. Instead of overpromising, Apple is pointing out potential problems with the latest AI models while exercising restraint. 

This content is part of EMARKETER’s subscription Briefings, where we pair daily updates with data and analysis from forecasts and research reports. Our Briefings prepare you to start your day informed, to provide critical insights in an important meeting, and to understand the context of what’s happening in your industry. Non-clients can click here to get a demo of our full platform and coverage.