Sora Video Generation, Gemini & Claude Updates, and AI in Healthcare Story

This week's edition covers OpenAI's Sora video launch, major updates from Google and Anthropic, Meta's video authentication tools, and AI's impact on healthcare documentation.

This Week in AI: Updates That Matter

Major Updates

🎬 Sora Launches with Sora Turbo: OpenAI's new video generator brings advanced capabilities.
The new Sora Turbo model enables video generation with tiered access levels. ChatGPT Plus users ($20/month) can create up to 50 videos monthly at 720p resolution with 5-second duration limits. Pro users ($200/month) get enhanced capabilities with up to 500 priority videos at 1080p resolution and 20-second duration, plus multiple simultaneous generations. Videos can be created in widescreen, vertical, or square formats. A new storyboard tool lets users precisely specify inputs for each frame.
Why it matters: This democratizes high-quality video creation while implementing essential safety measures like C2PA metadata, visible watermarks, and restrictions on uploads featuring people. Note: Service is unavailable in the EU, UK, and Switzerland.
openai.com
Explore Sora

🔒 Meta's Video Seal Advances AI Detection: New open-source framework enables invisible, resilient video watermarking.
The Video Seal framework adds imperceptible watermarks that survive common editing operations like blurring, cropping, and compression. Released under a permissive license, Meta is also launching Omni Seal Bench, a public leaderboard for neural watermarking research across different media types.
Why it matters: It provides crucial tools for verifying video authenticity as AI-generated content becomes more prevalent, particularly timely with Sora's launch.
ai.meta.com

🚀 Gemini 2.0 Brings Multimodal Output: Google's latest release adds sophisticated AI capabilities.
The update introduces Deep Research functionality for complex topic exploration, acting as an intelligent research assistant. The Flash experimental model is immediately available to all users, demonstrating improved reasoning abilities for sophisticated multi-step queries and multimodal interactions.
Why it matters: Enables developers to build more comprehensive AI applications that combine text, image, and audio interactions.
blog.google

Turns out, 'AI for everyone' was not the winning move

More AI everywhere doesn’t mean better results. Small teams of AI experts beat mass licenses. Attackers now use AI to run cybercrime at machine speed. Leaders can’t keep up. Tools still forget us. The winners? Focused expertise and hybrid AI, not AI everywhere.

Build AI · Dec 11 · 1 min

LLM-as-a-judge: the measurement problem

You've built something and you need to know if it works. So you do what's sensible—you ask an LLM to grade it. Factual accuracy, code quality, agent outputs. The machine judges the machine, and you get a number you can act on. Except that number

Understand AI · Dec 10 · 2 min

Tsinghua: focused AI expertise

Imagine you have a bunch of teams, some with AI, some without, and some where everyone gets their own AI. Researchers ran a big experiment with over 400 people to see what actually happens when you mix and match humans and AI in different ways. Here’s what they found: