OpenAI Launches GPT-5.6 Series with Government-Restricted Rollout

Title: Government Restrictions, Chip Independence Reshape AI Landscape

01OpenAI Limits GPT-5.6 Rollout After White House Request

OpenAI has begun a limited preview of its GPT-5.6 series, comprising three models: Sol as the flagship, Terra offering competitive performance to GPT-5.5 at half the cost, and Luna delivering strong capabilities at the lowest price point. The company had planned to make all three models generally available within weeks.

However, the rollout hit a wall when the White House requested a delay in public access. This follows a similar pattern set two weeks earlier when Anthropic had to take its most advanced AI models offline. OpenAI pushed back publicly, stating it "doesn't believe this kind of government access process should become the long-term default", arguing restrictions prevent developers, enterprises, and global partners from accessing critical tools.

02OpenAI Unveils Custom Jalapeño Chip, Joining Big Tech Exodus from Nvidia

OpenAI has revealed plans for Jalapeño, a custom inference chip built in partnership with Broadcom, marking a significant strategic shift away from Nvidia dependency. The move places OpenAI alongside Google, Apple, and SpaceX in a growing list of companies designing their own silicon to reduce single-supplier risk.

The chip development reflects a broader industry trend where AI companies seek greater control over their hardware stack. While Nvidia has dominated the AI chip market for years, the era of total dependence appears to be ending as major players invest in custom solutions tailored to their specific inference workloads. OpenAI's entry into custom silicon signals that even companies built on top of existing GPU infrastructure are now seeking vertical integration.

03Europe Pushes for Sovereign AI Capabilities Amid US Policy Uncertainty

European leaders are accelerating domestic AI development, with some explicitly citing Trump-era tensions as motivation for building independent AI infrastructure. The push aims to reduce reliance on US-based AI providers and establish continental control over critical AI capabilities.

The effort faces significant challenges—building a top-tier model competitive with OpenAI or Anthropic remains a stretch for European researchers and companies. However, the geopolitical friction with the current US administration has provided political momentum that previous European AI initiatives lacked. Whether Europe can close the capability gap remains uncertain, but the strategic imperative for sovereignty has never been stronger on the continent.

Also today

04
2,000 Hackers Failed to Leak Secrets from AI Assistant Challenge — Fernando Irarrázaval's OpenClaw security challenge attracted 6,000 attack attempts over email injection but no successful secret leaks after $500 in API costs.
05
OpenAI Hires Uber India Chief to Lead Its Largest Non-US Market — The strategic hire signals a major expansion push into India through new offices, partnerships, and local hiring.
06
AI Competition Has Moved Beyond Anthropic vs. OpenAI Binary — As AI capabilities carry real political consequences, analysts argue the industry dynamics now require collective action rather than pairwise competition.
07
OpenAI Internal Codex Usage Surged 13-56x Since November 2025 — OpenAI reports dramatic internal productivity gains with median output tokens growing 56x in Research, 32x in Customer Support, and 13x in Legal.
08
Anthropic Argues Its Own Success Is Necessary for AI Safety — Critics challenge Anthropic's accumulating power, but the company contends centralized control is essential for responsible AI development.
09
Hypothetical CVE-2026-LGTM Exposes AI Agent Cost Risks — A satirical incident report describes two competing AI review agents entering a disagreement loop costing $41,255 in inference before API keys were revoked.