Weekly AI News Update (12-18 December 2025)

Weekly AI News Update (12-18 December 2025)

Google is integrating its advanced Gemini AI model into Google Translate, significantly enhancing its text and speech translation capabilities. The update provides more natural and nuanced translations of idioms and local expressions, while a new beta feature enables real-time, speech-to-speech translation through headphones, preserving the speaker’s original tone and cadence. 🔗 Google
OpenAI has released a significantly upgraded image generation model, GPT-Image 1.5, within ChatGPT. The model features up to 4x faster generation, more precise and consistent editing capabilities, and improved rendering of text and complex details, offering users a more powerful tool for both practical edits and creative transformations. 🔗 OpenAI
The US government has launched the “U.S. Tech Force,” a two-year programme deploying approximately 1,000 engineers and specialists to advance AI infrastructure and digital projects across federal agencies. The initiative, backed by major technology companies, forms part of the administration’s broader strategy to centralise national AI policy and bolster the country’s competitive position. 🔗 CNBC
Google has released Gemini 3 Flash, a new AI model designed to deliver high reasoning performance at significantly faster speeds and lower costs. The model offers PhD-level benchmark scores and is optimised for agentic coding and development, becoming the new default in the Gemini app and Google Search to enhance user tasks with rapid, advanced intelligence. 🔗 Google
The Hong Kong Education Bureau has launched a funding programme to integrate artificial intelligence into local primary and secondary schools. Successful applicants will receive a one-off grant of HK$500,000 to purchase AI-powered devices or services and to subsidise student activities that enhance AI literacy and skills. 🔗 The Standard