News — 2026-09-24

The U.S. government has targeted critics of artificial intelligence, labeling them as "foreign agents" in a move that raises concerns about free speech and scientific debate. This action is seen as restrictive and could stifle open criticism of AI technologies.

The paper "COMED: The Missing Middle Between Routing and Collaboration in Multi-LLM Inference" introduces a new approach to improve the reliability of multi-Large-Language-Model inference systems by combining elements of both routing and collaboration methods. This is significant because it addresses limitations of existing strategies, offering potentially more robust performance across various queries.

arxiv.org↗

OpenAI launched its Academy program two years ago, aiming to expand access to artificial intelligence education across various communities. This initiative underscores the importance of democratizing AI knowledge and skills globally.

openai.com↗

Google Beam is expanding into five new countries and partnering with Industrious, broadening its global reach and network. This expansion enhances Google's presence in international markets and strengthens its offerings for businesses seeking global solutions.

Stripe has launched an AI platform aimed at providing customer support through knowledge articles, enhancing efficiency and reducing workload for support teams. This initiative is crucial as it leverages AI to improve service quality and scalability in the tech industry.

Ollama released version 0.34.4, improving structured outputs for thinking models and fixing issues like "model not found" errors and app responsiveness on macOS. These updates enhance reliability and user experience in the platform.

Radical Numerics employs advanced AI techniques, including biological chain-of-thought and multimodal perception, to advance bio-security. This approach helps in designing new genomes and gaining deeper insights into biology, crucial for keeping pace with the evolving bio-defense challenges.

Mercury 2.5, an advanced Large Language Model (LLM), achieved a processing speed of 770 tokens per second, marking a significant milestone in computational efficiency and capability within the AI sector. This performance improvement is crucial as it enhances the model's ability to handle complex tasks more swiftly and accurately, potentially revolutionizing applications ranging from customer service chatbots to advanced research tools.