Archive

Today's Key Insights

  • White House Seeks First Look at OpenAI, Anthropic Models — If adopted, the proposal would require OpenAI and Anthropic to delay sharing new models with the U.K.'s AI Safety Institute until U.S. agencies complete their review.
  • Oracle Sends Force Majeure Notice for New Mexico Stargate Site — If the New Mexico facility misses its 2028 go-live target, Oracle could delay payments under the force majeure notice.
  • Gemini’s Call for Me Calls Businesses on Pixel 11 — The first audience for Google’s business-calling feature is limited to U.S. Pixel 11 owners with paid Gemini subscriptions, leaving users outside that group without the stated access.
  • Pentagon Seeks $30.3M for AI Lie Detector — For the Department of Defense, the proposal would fund a five-year effort to improve lie-detector technology.
  • New Language Tool Flags Highest-Risk Suicide Cases — For people at the highest risk of suicide, the tool's stated purpose is to enable swifter interventions by identifying them from natural-language text.

Top Story

White House Seeks First Look at OpenAI, Anthropic Models #

The Decoder AI

The White House wants U.S. agencies to review new OpenAI and Anthropic models before British testers receive them. The Decoder AI reports that the administration is asking both companies to hold back models from the U.K.'s AI Safety Institute until that review happens.

The request would change the order of access: U.S. officials would examine the models first, followed by the British institute. The report does not identify a formal directive, review criteria, enforcement mechanism, or timeline for implementation.

That makes this an access-control proposal rather than a documented new testing regime. OpenAI and Anthropic could face delays in sharing models with British safety researchers, while the U.K. institute could lose early access to systems it expected to test.

Why it matters: If adopted, the proposal would require OpenAI and Anthropic to delay sharing new models with the U.K.'s AI Safety Institute until U.S. agencies complete their review.

Key Takeaways

  • The Decoder AI published the report on September 25, 2026.
  • The report does not specify review criteria or an implementation timeline.
  • It also does not identify a formal directive or enforcement mechanism.

Industry Updates

Oracle Sends Force Majeure Notice for New Mexico Stargate Site #

TechCrunch AI

Oracle has sent a force majeure notice tied to its New Mexico Stargate data center. The notice would allow Oracle to delay payments if the facility misses its target to come online in 2028.

Why it matters: If the New Mexico facility misses its 2028 go-live target, Oracle could delay payments under the force majeure notice.

Gemini’s Call for Me Calls Businesses on Pixel 11 #

TechCrunch AI • Wired AI

Google is testing an AI feature that lets Gemini call businesses for users. TechCrunch reports that the feature will first be available in the U.S. to Pixel 11 owners who pay for a Gemini subscription.

WIRED calls the feature “Call for Me” and says it is exclusive to the Pixel 11 series. WIRED also reports that Gemini can now make calls for users on Pixel phones.

Why it matters: The first audience for Google’s business-calling feature is limited to U.S. Pixel 11 owners with paid Gemini subscriptions, leaving users outside that group without the stated access.

Pentagon Seeks $30.3M for AI Lie Detector #

MIT Technology Review AI

The Pentagon wants $30.3 million over five years to develop an improved lie detector. The Department of Defense budget request calls the program “Polygraph+” or “Polygraph Next.”

The effort would focus on scoring algorithms that use artificial intelligence and machine learning, along with a technique called “standoff sensing.”

Why it matters: For the Department of Defense, the proposal would fund a five-year effort to improve lie-detector technology.

New Language Tool Flags Highest-Risk Suicide Cases #

MIT AI News

MIT AI News reports a new language-processing tool designed to identify individuals at the highest risk of suicide from natural-language text.

The tool's stated purpose is to enable swifter interventions by helping identify those highest-risk individuals.

Why it matters: For people at the highest risk of suicide, the tool's stated purpose is to enable swifter interventions by identifying them from natural-language text.