As the AI landscape continues to evolve rapidly, today’s news features groundbreaking developments in AI systems, alongside notable legal challenges faced by key players in the sector.
Meet ‘Code-as-World’: An Agentic Loop That Rewrites Real Videos Into Executable MuJoCo Physics Programs
A new approach called Code-as-World has been introduced, allowing for the extraction of editable MuJoCo scene code from real-world videos. This method not only enables the creation of accurate physical simulations but also serves as a training ground for AI systems to enhance their physical reasoning capabilities.
Sony Music, Warner sue Anthropic, alleging a “brazen campaign” of intellectual property theft
In a significant legal move, Sony Music and Warner have filed a lawsuit against Anthropic, accusing the company of engaging in extensive intellectual property theft. The lawsuit highlights the growing concerns over the protection of creative works in the age of AI, as the music industry grapples with the implications of AI-generated content.
Google AI Releases Gemini Omni 1.1 Flash: 40-Second Scene Extension, First/Last Frame Control, and 4K Upscaling
Google has rolled out an update for its Gemini Omni model, introducing features that enhance video generation and editing capabilities. The new version allows for longer scene extensions, improved control over camera movements, and the ability to upscale videos to 4K, showcasing advancements in multimodal AI technology.
AI-generated videos are already displacing actors and livestreamers across China’s entertainment industry
The rise of AI-generated content in China is having a profound impact on the entertainment sector, with a staggering 95% of short dramas produced in early 2026 being created by AI. This shift is leading to significant job losses for actors and livestreamers, prompting concerns over labor rights and the future of creative professions in the face of advancing technology.
Google’s WikiSkill gives AI agents a persistent memory of past mistakes to sharpen future performance
Google Research has unveiled WikiSkill, a new framework that enables AI agents to retain and learn from their past experiences. By documenting both successes and failures in a structured format, these agents can improve their performance over time, making it a promising development for enhancing AI capabilities.
LAION drops massive open video dataset with 10 million hours of footage for AI research
LAION has released one of the largest open video datasets available for AI research, featuring 10 million hours of footage. This dataset is expected to significantly enhance the training of AI models, surpassing previous benchmarks and providing researchers with valuable resources for developing more sophisticated AI applications.
Anthropic wants to do for physical hardware what its Model Context Protocol did for software
Anthropic is working on a new initiative called the Model Hardware Standard, aimed at streamlining the integration of AI systems with physical devices. Early tests indicate that this framework could dramatically reduce the time required for integration, although challenges remain in ensuring that AI understands physical interactions correctly.
Google Deepmind’s AI Co-Scientist now plans experiments, runs lab equipment, and writes scientific papers
Google Deepmind has advanced its AI Co-Scientist, which now plays a more active role in scientific research by planning experiments and operating lab equipment. This development marks a significant step in integrating AI into the research process, with the system demonstrating its ability to generate valid experimental results across various scientific disciplines.
U.S. court rules Pentagon’s blacklisting of Anthropic was unlawful
A federal court has determined that the Pentagon’s decision to blacklist Anthropic was unlawful, highlighting the ongoing tensions between government policies and AI companies. This ruling may have implications for how the Department of Defense interacts with AI firms, particularly those that critique government practices.
Compiled automatically by the Tech AI Newsdesk from public AI-news sources and summarised in our own words.
