Top topics
Find news and summaries about the products and technologies most covered in public news.
News (page 147)
In macOS 27, users report that Siri continues running even after being disabled via Settings or Screen Time, prompting discussion.
- •Even when Siri-related services are disabled through the Settings app or third‑party tools, the “Siri AI.app” process appears in Activity Monitor.
- •This behavior resembles Microsoft’s Recall feature and Apple’s Personal Context feature, raising privacy concerns.
- •It suggests that AI functions are forcibly implemented in macOS 27 and that users cannot fully disable them.
A tool called “Lossless-memory” that provides a personal AI memory function that retains information without summarizing has been announced.
Meta's AI agent Muse was reported to have its access blocked on Amazon for violating terms of service when used to make purchases on behalf of users.
- •On Sunday, Muse users began seeing a pop‑up message stating that access by unauthorized AI agents violates the terms of service.
- •According to GeekWire, Meta had not previously notified Amazon about the use of its AI agent.
- •The action was taken based on Amazon's terms of service, which customers have agreed to.
The article critiques reliance on AI for writing and emphasizes preserving human originality.
In September, AI-generated code accounted for 17.25% of all Linux kernel patches submitted.
A small-scale decision-making model family called “Kev,” similar to Jev and built on Qwen3.5, has been introduced.
The dynamic continual‑learning model “Mini‑AGI”, which runs on an 8 GB VRAM setup, has been released and its scaling‑law graph is reported as promising.
- •It is claimed to incorporate every element the authors wanted to include in an AGI model.
- •A sample training run and the scaling‑law graph are available in the GitHub repository.
- •Training on the relatively modest 8 GB VRAM resource is highlighted as a technical feature.
Heretic is gaining attention as a tool or method for removing restrictions on language models.
AI chatbots frequently fail to provide accurate information for finance-related inquiries, often giving incorrect answers.
The system architecture and design philosophy consisting of a deterministic core and a nondeterministic shell are being discussed.