Zero-shot learning enables an AI model to handle a task or category without receiving a labeled example of that exact case during task-specific training. The model relies on broader representations, descriptions, instructions, or relationships learned earlier. A language model may classify text from a written label definition, while a vision system can connect unseen classes to semantic attributes. This flexibility is useful when categories change quickly or examples are scarce. Performance usually trails well-trained supervised systems and depends on whether the new task is represented in the model’s prior knowledge. Clear instructions, meaningful labels, confidence thresholds, and evaluation on truly unseen cases help distinguish genuine generalization from accidental familiarity with training data.
Chinese generative AI startups MiniMax and Zhipu AI have disclosed their financials ahead of potential Hong Kong listings, highlighting modest revenues and mounting losses compared with U.S. peers.
Adobe’s Firefly app adds a new video editor supporting precise prompt-based edits, along with integration of third-party models including Black Forest Labs’ FLUX.2 and Topaz Astra for enhanced video and image generation.
Zoom introduced AI Companion 3.0, a federated AI solution integrating internal and third-party models to streamline meetings, documents, and personal workflows. The update adds agentic AI features, a web interface, and enhanced productivity tools.
Nvidia is reportedly planning to increase production of its H200 AI training chips following approval to sell them in China.
OpenAI faces a wrongful death lawsuit alleging ChatGPT reinforced a man’s violent delusions, contributing to the killing of his mother in Connecticut. The case marks the first time an AI platform has been accused of involvement in a murder.
Google introduced the Titans architecture and the MIRAS framework to enable AI models to handle massive contexts and update their internal memory while running, improving performance in long-sequence tasks.
ChatGPT now reaches more than 800 million weekly users, creating a powerful adoption flywheel that is speeding the transition from AI experimentation to full-scale enterprise deployment, according to OpenAI’s new 2025 report.
OpenAI buys Neptune for under $400 million in stock, gaining advanced AI model tracking tools and strengthening capabilities in large-scale AI development.
OpenAI introduces an experimental “confessions” system that rewards honest self-reporting of mistakes in GPT-5 Thinking models, aiming to improve AI safety and visibility of failures.
AWS introduced new tools for enterprise customers to build and fine-tune custom large language models, alongside AI infrastructure offerings for on-premises deployments.