AI News Briefing

Following the reported White House request, here's the official announcement, OpenAI's official preview introduces GPT-5.6 Sol, a next-generation flagship with improved coding, science, and cybersecurity capabilities paired with its most advanced safety stack. It is the primary-source announcement of the model.

Continuing our coverage from yesterday, OpenAI announced a limited preview of GPT-5.6 after a US government request to stagger its release, explicitly criticizing the move as keeping the best tools from users and cyber defenders. The action echoes the recent handling of Anthropic's Mythos product.

Continuing our coverage from yesterday, OpenAI limited GPT-5.6 availability following a government request, stating that customer-by-customer access approval should not become the long-term default. The company framed the restriction as harmful to developers, enterprises, and cyber defenders.

The New York Times moved to amend its copyright complaint to allege Microsoft actively induced OpenAI's alleged infringement by building a bespoke supercomputer ranked among the world's most powerful. The amendment responds to a recent Supreme Court ruling in the Cox Communications case that raised the bar for contributory infringement, now requiring proof of intentional inducement.

Building on the Jalapeño chip announcement, A video segment examines the broad industry shift toward custom AI silicon, highlighting OpenAI's Jalapeño inference chip built with Broadcom alongside efforts from Google, Apple, and SpaceX. The trend reflects efforts to reduce single-supplier dependence on Nvidia.