OpenAI is expanding ChatGPT’s advertising tools with Sponsored Agents, a format that opens a separate chat with an AI representative of the brand after an ad is clicked. The feature is currently in a limited alpha test in the US.
OpenAI has introduced a framework for documenting and disclosing instances of model misalignment. Alongside it, the company published six internal reports from training and evaluations.
Salesforce developed Koa with NVIDIA for multi-step CRM tasks in Agentforce. The company post-trained it on synthetic scenarios and says it did not use customer data.
Google has introduced Gemini 3.8 Live and Gemini 3.8 Live Extended Thinking. They are designed for voice applications with image, video, and text inputs.
OpenAI says in a customer study that Perplexity uses GPT‑6 Astra for communications, system changes, workflow testing, and production software monitoring. The model’s permissions were not disclosed.
Apple now allows private interactions to be used to improve foundation models, but only with the user’s explicit, voluntary consent.
Sam Altman ruled out OpenAI going public in 2026 in an interview with Fortune. The company, he said, does not feel pressure to pursue an IPO and is focused on AI safety and alignment.
Anthropic describes a group from northern Yemen that reportedly used Claude Code to develop guidance, navigation and flight-control software. The company blocked the accounts but has no evidence that a functional weapon was deployed.
On September 10, 2026, OpenAI announced the public beta of the Agents API. The interface is intended to provide developers with a managed layer for longer-running cloud agents, including tools, sandboxes, and subagent coordination.
NVIDIA announced cooperation with eight Australian companies on AI data center projects with a planned combined capacity of up to 2 GW by 2027.