AI news

1. EU AI Office Issues First 'Systemic Risk' Designations Under AI Act

Date: September 1, 2026

The European AI Office has officially designated four frontier AI models as posing "Systemic Risk" under the full enforcement phase of the EU AI Act. This designation mandates that the developers—OpenAI, Google, Meta, and Anthropic—must now perform mandatory adversarial testing, implement enhanced cybersecurity protocols, and provide detailed reports on the energy consumption of their models to the European Commission. This marks the transition from voluntary compliance to active regulatory enforcement for General Purpose AI (GPAI) models within the Union.

Key Facts & Business Impact:

Requirement Deadline Penalty for Non--Compliance
Adversarial Testing Results Oct 15, 2026 Up to 7% of global turnover
Energy Efficiency Report Dec 01, 2026 Up to 3% of global turnover

Source: https://www.reuters.com/technology/eu--ai--act--enforcement--systemic--risk--2026--09--01/