Meta introduces an on-device scam detection feature for WhatsApp that identifies suspicious messages locally, preserving end-to-end encryption while balancing privacy and safety.
Nilay Patel argues that continuous recording and cloud processing are inherent to AR glasses, making privacy invasion inevitable, and questions whether such products should exist at all.
Meta shares a hybrid asset classification approach: using LLMs for ambiguous cold-start but relying on human-reviewed deterministic rules for daily enforcement, achieving auditable data governance in the AI era.
Deep research agents combining internal and web data leak secrets through query logs; a new benchmark and privacy-aware RL training provide metrics and solutions.
Apple's new Siri AI at WWDC 2026 adopts a more pragmatic approach by integrating a custom Gemini model, vision LLMs, and the Core AI library. Simon Willison remains cautiously optimistic, stressing he'll believe it when he sees it.
The FTC fined three companies for falsely advertising an AI 'active listening' service that was actually just reselling email lists. This case confirms that the 'phone spying for ads' conspiracy is often a marketing scam, not real technology.
Onco智能体 proposes a dual-tier multi-agent framework that provides reliable and secure AI assistance for oncology clinical decisions through localized deployment, tiered models, and strict privacy protection.
Simon Willison adapted LlamaIndex's LiteParse into a pure browser-based version, enabling local PDF text extraction and OCR without a server, highlighting privacy and the importance of spatial text parsing.
Meta is strengthening end-to-end encrypted backups for WhatsApp and Messenger using hardware security modules (HSMs) and transparent deployment, making backups inaccessible even to Meta itself, with independent third-party audits.