🎬 ByteDance previews Seedance 2.0 video model, boosting China AI stocks 🎬
New AI video model generates 2K cinematic content 30% faster than its predecessor.

ByteDance has launched a pre-release version of Seedance 2.0 on its Jimeng AI platform, triggering a rally in Chinese AI and entertainment stocks. The model supports text, image, video, and audio inputs to generate what ByteDance calls "cinematic content" with 2K resolution and speeds 30% faster than its predecessor. A standout feature is multi-lens storytelling, where a single prompt expands into multiple connected scenes with consistent characters, lighting, and tone throughout. Swiss consultancy CTOL described Seedance 2.0 as the "most advanced AI video generation model available," surpassing OpenAI's Sora 2 and Google's Veo 3.1 in practical testing. The launch pushed COL Group to its daily 20% price limit while Shanghai Film and Perfect World rose about 10%.

💰 ChatGPT rolls out ads for US Free users 💰
OpenAI begins monetizing its chatbot with targeted advertising for non-paying users.

OpenAI has officially begun testing ads inside ChatGPT for US users on its Free and $8/month Go plans. Ads appear below chat replies and are targeted using conversation context, past chats, and ad engagement, though OpenAI insists ads won't influence the model's answers. Advertisers cannot access user chats or personal data, and ads won't appear around sensitive topics like health or politics. The move drew ridicule from Anthropic in Super Bowl ads that mocked poorly integrated AI advertising, prompting CEO Sam Altman to call the spots "dishonest." This revenue diversification signals OpenAI's push to cover rising development costs as competition intensifies.

🚗 Waymo builds driving simulator on Google's Genie 3 🚗
Google's autonomous driving unit unveils AI-powered tool for hyper-realistic robotaxi training.

Waymo has introduced the Waymo World Model, a simulation system built on Google DeepMind's Genie 3 architecture to create hyper-realistic driving scenarios for robotaxi training. The model generates high-fidelity camera and lidar data simultaneously, allowing engineers to simulate rare events like tornadoes, elephant encounters, or reckless drivers through simple language prompts. Strong controllability enables "what if" counterfactual testing, weather mutations, and time-of-day changes. The system can even convert regular dashcam footage into multimodal simulations, offering new ways to expand training data as Waymo scales its service to new cities.

INTERESTING TO KNOW

🤝 Anthropic closing in on $20B funding round 🤝

Anthropic is closing in on a $20 billion funding round that would push the company's valuation to $350 billion, according to Bloomberg. The massive capital injection comes as Anthropic competes directly with OpenAI following recent product launches and a high-profile Super Bowl advertising campaign that mocked rival approaches to monetization.

💻 Cursor releases Composer 1.5 coding model 💻

Cursor has released Composer 1.5, its updated AI coding assistant built by scaling reinforcement learning 20x further on the same pretrained base. The model responds quickly on easy problems while reasoning longer on hard ones, and features self-summarization to maintain accuracy across extended contexts. Cursor says this training approach demonstrates that reinforcement learning for coding can be continually scaled with predictable intelligence improvements.