Source: TechCrunch (Dec  16, 2025)

Adobe has expanded its Firefly AI platform with a new video editor that enables precise text-based editing of existing clips, moving beyond its previous prompt-only generation capabilities. The editor, which left private beta in October, allows users to modify video elements, adjust colors and camera angles through natural language commands, and includes a timeline view for frame-level adjustments.

The update integrates multiple third-party AI models: Runway's Aleph powers the text-based editing features, Black Forest Labs' FLUX.2 handles image generation (launching across Firefly immediately and in Adobe Express from January), and Topaz Labs' Astra enables video upscaling to 1080p or 4K resolution. Adobe's own Firefly Video model now supports camera motion replication using reference videos.

To encourage adoption amid growing competition, Adobe is offering unlimited generations from all image models and its video model through January 15 for subscribers of Firefly Pro, Premium, and higher-tier credit plans. The company also introduced collaborative boards and continues its strategy of integrating various AI models into its platform.

Source: The Verge (Dec  16, 2025)

OpenAI launched GPT Image 1.5 on December 16, 2025, as its new flagship image generation model, bringing significant improvements to ChatGPT's image capabilities. The model delivers generation speeds up to 4 times faster than before OpenAI while maintaining enhanced quality and precision.

Key improvements include stronger instruction following and more precise editing that changes only what users request while keeping elements like lighting, composition, and people's appearance consistent. The model also features improved text rendering to handle denser and smaller text more accurately.

OpenAI introduced a dedicated Images section in ChatGPT's sidebar that functions more like a creative studio with preset styles and trending prompts. The model is available to all ChatGPT users globally and through the API as GPT-Image-1.5.

The release represents the latest escalation in competition with Google's Gemini, following an internal memo from OpenAI CEO Sam Altman declaring a "code red" to, as the company works to maintain its position against Google's recent AI advances.

Source: Psychiatrictimes (Dec 16, 2025)

AI chatbots are collecting vast amounts of personal data with minimal privacy protections, according to a new analysis in Psychiatric Times. Major platforms like ChatGPT and Google's Gemini use "opt-out" systems that automatically grant permission to use conversation data for model training unless users manually disable it.

Recent security incidents underscore the risks. In 2025, hackers used Claude to extort up to $500,000 from 17 organizations across healthcare and government sectors. Italy fined OpenAI 15 million euros in 2024 for lacking transparency in data handling, and a U.S. court order now requires OpenAI to retain all conversations indefinitely — including those users attempted to delete.

Experts warn that aggregated chatbot data enables targeted manipulation and behavioral prediction, similar to how Cambridge Analytica used quiz data from 87 million Facebook users to influence the 2016 election. The analysis concludes that meaningful chatbot privacy doesn't currently exist and urges users to exercise caution when sharing sensitive information until stronger legal protections are established.

Your weekly must-try AI tools & resources

The Challenge: Use AI to design your perfect week-long digital detox. Have Claude create a day-by-day plan that includes offline activities, analog hobbies to explore, nature experiences, and creative projects. The twist? Use AI as your planning partner, then go tech-minimal for the actual experience.

Bonus points: Share one photo from your detox week (no captions needed) or journal about your favorite offline moment.

Catch you next Wednesday,