Good morning. It’s Friday, December 13th.

Did you know: On this day in 1980, Apple went public. The initial stock price was $22 per share ($0.10 if adjusted for the 5 stock splits that have happened since then).

In today’s email:

• OpenAI’s Live Video Released

• Google’s AI Agent

• 5 New AI Tools

• Latest AI Research Papers

You read. We listen. Let us know what you think by replying to this email.

OpenAI Introduces Live Video, Screen Sharing, and 'Santa Mode' as Shipmas Rollout Continues

For Day 5, Apple added ChatGPT to its Intelligence features in iOS 18.2 and macOS Sequoia, embedding it within its Apple Intelligence ecosystem. The chatbot collaborates with Siri, stepping in for tasks where its capabilities surpass the assistant’s. Users retain full control, with privacy safeguards ensuring no data is stored unless they sign into an OpenAI account, in which case OpenAI’s privacy policies apply.

This update arrives alongside other AI features such as Image Playground for creative photo editing and Writing Tools for text composition. While these innovations enhance Apple’s operating systems, many announced features remain delayed, with broader rollouts anticipated by 2025. Regulatory challenges in the EU have further restricted availability.

For Day 6 OpenAI added live video and screen sharing to Advanced Voice Mode, now accessible to Teams, Plus, and Pro users on iOS and Android. These features empower the AI to process live visual input, recognize objects, and offer context-aware guidance, as demonstrated in the "12 Days of Shipmas" event, where ChatGPT assisted users with tasks like brewing coffee by analyzing their on-screen environment.

The integration of screen-sharing expands its utility, enabling real-time interaction with browser-based content, allowing users to seek insights on apps or images. These updates align with similar innovations from Google’s Project Astra and Microsoft’s Copilot Vision, marking a broader trend of embedding AI into everyday workflows. Additionally, OpenAI has rolled out a festive "Santa Mode," blending holiday cheer with advanced voice technology across platforms. Read more.

Google unveils Project Mariner: AI agents to use the web for you

Google’s Project Mariner introduces a groundbreaking AI agent that autonomously interacts with websites, simulating human browsing actions within Chrome. Powered by Gemini, the agent navigates, clicks, and fills out forms, providing users with an entirely passive web experience. Initially available to a select group, it can handle tasks such as shopping, flight booking, and information retrieval.

However, the agent cannot process payments or accept terms, ensuring user control. Restricted to the active browser tab, users must remain engaged while it works. Read more.

• Google’s new Jules AI agent will help developers fix buggy code

• How Google's Project Astra gives an AI agent eyes

• ‘Not there yet’: Sora rollout receives mixed response from AI filmmakers citing inconsistent results, content restrictions

• Google’s new Trillium AI chip delivers 4x speed and powers Gemini 2.0

• ​​Scaling Smarts, Not Size: Microsoft’s Phi-4 Reshapes AI Priorities

• Google launches Gemini 2.0, focusing on AI agents and multimodal capabilities

• Anthropic’s Claude 3.5 Haiku Challenges OpenAI’s Dominance in AI Coding

• Apple develops its own AI chip "Baltra"

• Microsoft AI’s CEO poaches former Google DeepMind colleagues for a new AI health unit

• The AI boom nets hundreds of Australian data center employees a $41,300 holiday bonus

• ChatGPT and Sora experienced a major outage

• Meta’s AI Leap: Motivo and LCM Redefine Virtual Interactions

• Gemini AI can now summarize what’s in your Google Drive folders

• Twelve Labs is building AI that can analyze and search through videos

• Harvard Is Releasing a Massive Free AI Training Dataset Funded by OpenAI and Microsoft

• Midjourney launches Patchwork collaborative canvas tool for visual storytelling

• Google built an AI tool that can do research for you

• Systematic use of GenAI cognitively overwhelms many people, Microsoft Research finds

• Study: OpenAI's o1 relies on trial-and-error and informal reasoning

• Meta debuts a tool for watermarking AI-generated videos

• LMSYS finally adds an image gen leaderboard and arena

📄 SynCamMaster: Synchronizing Multi-Camera Video Generation from Diverse Viewpoints

📄 LAION-SG: An Enhanced Large-Scale Dataset for Training Complex Image-Text Models with Structural Annotations ​​

📄 POINTS1.5: Building a Vision-Language Model towards Real World Applications

📄 Euclid: Supercharging Multimodal LLMs with Synthetic High-Fidelity Visual Descriptions

📄 InternLM-XComposer2.5-OmniLive: A Comprehensive Multimodal System for Long-term Streaming Video and Audio Interactions

Thank you for reading today’s edition.

Your feedback is valuable. Respond to this email and tell us how you think we could add more value to this newsletter.

Interested in reaching smart readers like you? To become an AI Breakfast sponsor, reply to this email or DM us on X!