Meta plans to use browsing history from other sites for feed and AI response personalization, according to ITmedia NEWS (@itmedia_news). While the goal is to improve ad/recommendation systems and AI assistant quality, cross-site tracking and privacy issues could resurface.
A re-analysis of historical AI model performance charts suggests that actual performance gains may be significantly larger than officially reported, with some models showing distinct patterns that deviate from standard curve fitting. Separately, Google has announced its Gemini 3.5 Live Translate speech model, aiming for near real-time interpretation in applications like Google Translate. Meanwhile, Meta plans to leverage cross-site user behavior data for personalizing feeds and AI responses, raising potential privacy concerns. AI
IMPACT Google's real-time translation model could improve multilingual communication, while Meta's data usage plans raise privacy questions for AI personalization.