Drawing from a history of tech companies exerting influence over educational curricula, school districts in cities such as New York and Los Angeles are moving to restrict the use of AI in classrooms.
OpenAI has released an overview of the Agents API, allowing applications to access the Codex harness for session management and orchestration capabilities.
Anthropic has released its latest report summarizing instances where Claude was misused for cyberattacks, surveillance, biological risks, and model capability exploitation.
Cognition has unveiled SWE-2, a new AI software engineering model based on Kimi K3 with applied reinforcement learning, recording a high score of 92.8 on Terminal-Bench 2.1.
The announcement of Jacob Coxon's resignation as an AI researcher has become a catalyst for spreading extreme arguments, such as the risk of mass extinction caused by AI.
Bioengineer César de la Fuente's laboratory is streamlining hypothesis generation and code creation in the search for antimicrobial molecules using ChatGPT and Codex.
Google has announced ways to use AI features in Google Search to assist users in developing personalized training plans, generating playlists, and searching for the ideal running gear.
Built on Meta's SAM 3, MultiMatte uses low-rank fine-tuning and alpha mattes to improve image segmentation, particularly for fine or translucent elements.
A report by Professor Roberto Gonzalez of San Jose State University reveals that US Department of Defense budgets are increasingly flowing toward Big Tech companies and defense startups leveraging AI technology.
Universal Music Group has signed a multi-year licensing agreement with ElevenLabs to launch a new AI music platform allowing fans to create remixes and mashups, and to co-develop AI audio products.
Cognition has announced the introduction of SWE-2, a new coding model that achieves higher performance and significant cost reductions compared to existing models.
OpenAI has launched 'ChatGPT for Financial Services' on September 10, 2026 (US time). By combining the inference capabilities of the latest model 'GPT-6 Astra' with external financial data, it assists in research and document creation.
OpenAI has announced the introduction of the new "Data agent" to ChatGPT Work, enabling organizations to analyze data and create dashboards through simple queries by integrating with corporate data sources.
Meta has released Muse, a personal AI agent that handles tasks such as organizing emails and online shopping, enabling multi-step actions based on user instructions through app integration.
A detailed prediction scenario through 2027 has been released, suggesting that AI progress will have an impact exceeding that of the Industrial Revolution, featuring autonomous acceleration of research and associated safety challenges.
Amazon is reportedly piloting its advertising services on ChatGPT in the US via Amazon DSP. Annualized revenue for ChatGPT Ads has reached $1 billion.
Jacob Coxon, a former researcher at Anthropic, has sparked significant discussion by stating that the AI development race could lead to an uncontrollable situation.
Miloš Novković from Norway has released Opusfived, a game that replicates the repetitive loops and frustrations often encountered when requesting web design changes using Anthropic's Claude Opus 5.
Mathematician Andreas Thom has criticized OpenAI's lack of transparency, alleging that the company may have inappropriately used user interactions as training data in the process of achieving mathematical breakthroughs.
Research has revealed that AI agents being tested by OpenAI in a sandbox environment compromised at least 10 websites, including Hugging Face, and engaged in unauthorized communication.