Top Stories – July 25, 2026
1. EO 14409’s Voluntary AI Review Framework Is Seven Days From Its Defining Moment
Source: TechTimes | Date: July 24, 2026
New analysis warns that EO 14409’s “voluntary” pre-release framework is effectively binding: NSA, CISA, and NIST classified benchmarks defining which developers face government review are due August 1, while the White House has already demonstrated it will use parallel export-control authority to suspend models entirely outside EO 14409’s scope.
2. EU Digital Omnibus Must Clear Official Journal by Monday to Defer August 2 High-Risk AI Deadline
Source: DLA Piper | Date: July 25, 2026
With August 2 enforcement eight days away, the EU Digital Omnibus – which would defer Annex III high-risk AI obligations from August 2026 to December 2027 – still has not appeared in the Official Journal; publication must occur by approximately July 30 for the three-day entry-into-force period to clear before the deadline.
3. China MOFCOM Gives No Timeline on AI Model Weight Export Controls
Source: TechTimes | Date: July 22, 2026
Four days after Reuters and the Financial Times reported China’s Ministry of Commerce was weighing tiered controls on AI model weights for Qwen, Doubao, GLM-5.2, and DeepSeek R1, Beijing has announced no final decision or timetable, with affected companies arguing restrictions would slow domestic AI development.