Regulation (EU) 2024/1689 sets cybersecurity and accuracy expectations for high-risk AI systems. Article 15 covers accuracy, resilience, and cybersecurity, including resistance to attempts that alter use or performance. Article 9 requires a risk-management system, and that system is supposed to include testing. For many SaaS teams shipping LLM features, the practical question is not the legal text alone. It is what evidence you can hand to a customer, investor, or internal compliance lead when they ask how you tested against adversarial abuse.
The Digital Omnibus on AI moved several application dates.
A timeline maintained by the Future of Life Institute, updated 31 August 2026 from the amended text, puts Chapter III (including Articles 9 and 15) for Annex III high-risk systems at 2 December 2027, and for Annex I product systems at 2 August 2028. Article 50 transparency obligations applied on 2 August 2026. Annex III rules are not in force yet. Planning still matters because technical documentation and testing habits take time to build.
September 11 is coming up fast, and I've had the same conversation with four different founders this month: "does the Cyber Resilience Act apply to us?" Every one of them assumed yes. Two of them were wrong.
I run penetration testing and AI red teaming for SaaS companies at Faultline Security, so this is squarely in my lane, and I wanted to write the honest version of this instead of the vendor version, because the vendor version says "everyone is in scope" and that's just not accurate.
What's actually happening on September 11
The CRA's reporting obligations become enforceable. If you're a manufacturer of a "product with digital elements," you now have to report actively exploited vulnerabilities and severe security incidents through a single platform, on a real clock: 24 hours for the early warning, 72 hours for a fuller notification, then a final report within 14 days (vulnerabilities) or a month (severe incidents). Fines for getting reporting wrong run up to 15M or 2.5% of global turnover.
Microsoft announced two new security products at a small San Francisco event yesterday: MAI-Cyber-1-Flash, the company's first cybersecurity-specialized AI model, and Perception, an agentic security platform that deploys teams of AI agents to find and fix vulnerabilities.
The preview opens November 3. That's the useful clock.
What the two products do
MAI-Cyber-1-Flash is built specifically for finding vulnerabilities in complex codebases. It powers MDASH, Microsoft's harness for software vulnerability identification and remediation.
