Apple Clarifies Gemini Rumors: Siri AI to Be Built In-House
Following the 2026 Worldwide Developers Conference (WWDC), Apple addressed media inquiries regarding rumors that its new AI infrastructure was merely a "Gemini shell." The company clarified that Apple Foundation Models (AFM) are not a rebranding of Google’s Gemini technology but are instead fully developed in-house and operate independently.

Apple explained that while the AFM models leveraged Google’s Gemini technology during the distillation and training phases, the final output relies entirely on Apple’s proprietary code, infrastructure, and datasets. Consequently, users do not interact directly with Google’s source code or Gemini intelligence, nor does the system connect to Google services.
The Apple Foundation Models ecosystem currently comprises five distinct models. The edge-focused AFM Core handles fundamental AI tasks, while AF Core Advanced employs a sparse architecture with native multimodal capabilities to support complex local operations. For cloud-based demands, AFM Cloud manages high-volume requests, M Cloud Image specializes in image generation and editing, and AFM Cloud Pro is engineered for agent-based tools and intensive workloads. Each model is optimized for Apple Silicon and trained on proprietary data, with distillation assistance from Gemini.
Apple further noted that although its cloud models utilize Google Cloud servers and NVIDIA graphics processing units, they maintain Private Cloud Compute certification to guarantee user data security and privacy compliance.
Related article
OpenAI Still Struggles to Control Rogue AI Activity
On Friday, OpenAI launched a dedicated page for “misalignment reports,” revealing an alarming volume of rogue behaviors documented over an extended period. The site currently lists nine incidents, the majority occurring during reinforcement learning
Flock CEO claims mics detect only gunshots despite website claims
Flock CEO Garrett Langley stated that the company’s gunshot-detection microphones capture “only gunshots,” yet Flock’s website presents a much wider scope of the ambient audio it records.Speaking at TechCrunch Disrupt, the surveillance firm’s co-foun
Alibaba Unveils Qwen3.5-Omni, Delivering Top-Tier Cross-Modal Performance
China’s leading AI models are rapidly evolving from followers to pioneers in multimodal interaction.On March 30, Alibaba unveiled the next-generation multimodal large model, Qwen3.5-Omni. Leveraging a hybrid attention MoE architecture, this model ena
Related Special Topic Recommendations
Comments (0)
0/500
Following the 2026 Worldwide Developers Conference (WWDC), Apple addressed media inquiries regarding rumors that its new AI infrastructure was merely a "Gemini shell." The company clarified that Apple Foundation Models (AFM) are not a rebranding of Google’s Gemini technology but are instead fully developed in-house and operate independently.

Apple explained that while the AFM models leveraged Google’s Gemini technology during the distillation and training phases, the final output relies entirely on Apple’s proprietary code, infrastructure, and datasets. Consequently, users do not interact directly with Google’s source code or Gemini intelligence, nor does the system connect to Google services.
The Apple Foundation Models ecosystem currently comprises five distinct models. The edge-focused AFM Core handles fundamental AI tasks, while AF Core Advanced employs a sparse architecture with native multimodal capabilities to support complex local operations. For cloud-based demands, AFM Cloud manages high-volume requests, M Cloud Image specializes in image generation and editing, and AFM Cloud Pro is engineered for agent-based tools and intensive workloads. Each model is optimized for Apple Silicon and trained on proprietary data, with distillation assistance from Gemini.
Apple further noted that although its cloud models utilize Google Cloud servers and NVIDIA graphics processing units, they maintain Private Cloud Compute certification to guarantee user data security and privacy compliance.
OpenAI Still Struggles to Control Rogue AI Activity
On Friday, OpenAI launched a dedicated page for “misalignment reports,” revealing an alarming volume of rogue behaviors documented over an extended period. The site currently lists nine incidents, the majority occurring during reinforcement learning
Flock CEO claims mics detect only gunshots despite website claims
Flock CEO Garrett Langley stated that the company’s gunshot-detection microphones capture “only gunshots,” yet Flock’s website presents a much wider scope of the ambient audio it records.Speaking at TechCrunch Disrupt, the surveillance firm’s co-foun
Alibaba Unveils Qwen3.5-Omni, Delivering Top-Tier Cross-Modal Performance
China’s leading AI models are rapidly evolving from followers to pioneers in multimodal interaction.On March 30, Alibaba unveiled the next-generation multimodal large model, Qwen3.5-Omni. Leveraging a hybrid attention MoE architecture, this model ena





Home






