Google DeepMind CEO Demis Hassabis is pushing for a US-led international AI watchdog with the authority to evaluate frontier models and halt deployments deemed too dangerous, potentially before the end of 2025.
Sixty years after ELIZA became the world's first chatbot, recovered source code reveals how its design questions still haunt today's AI industry.
Singapore-based PixVerse has secured $439 million in its Series C round, with backing from Alibaba and others, pushing its valuation past $2 billion as it expands globally.
Anthropic has discovered a hidden internal space called the J-space inside its Claude AI model, where unseen words appear to influence how the model reasons through problems.
Lyzr deployed its own AI agent to manage a $100 million Series B raise, proving its product works while pulling $400 million in investor interest.
A Brown University professor's decision to offer take-home exams led to unprecedented perfect scores, exposing what appears to be widespread AI-assisted cheating among Ivy League students.
OpenAI has started a limited preview of its next-generation GPT-5.6 model series, introducing Sol, Terra, and Luna with enhanced safety, agentic capabilities, and tiered pricing.
Prime Intellect has raised $130 million at a $1 billion valuation, offering companies a full-stack platform to build and train their own AI agents without depending on frontier AI labs.
ZML, a Paris-based AI startup backed by Yann LeCun, has released a free LLM inference server designed to run open-source models across multiple chip platforms at peak performance.
LeRobot's latest release introduces world model policies that learn to imagine the future, a growing zoo of vision-language-action models, new reward tools and six evaluation benchmarks.