What the Glasswing Narrative Erases: A substrate-layer response to the software quality reframe of cybersecurity

Project Glasswing has been framed by prominent voices as evidence that cybersecurity's future lies in AI-assisted vulnerability remediation—that the discipline's core problem is software quality, and that sufficiently capable AI models will address that problem at scale. This analysis argues that framing collapses cybersecurity into a single subdomain, erases the adversarial, sociotechnical, and governance dimensions of the field, and misrepresents the actual threat surface that Mythos-class models introduce. The paper identifies four structural elisions in the software quality narrative and enumerates the security domains that do not disappear when software quality improves. It concludes that the substrate-layer governance requirements introduced by AI agent deployment represent an entirely new security surface that the Glasswing narrative does not address and may actively obscure. Related prior work: The Deeper Governance Implications of Mythos-Class Models: A Substrate-Layer Analysis. DOI 10.5281/zenodo.19473539. Never Why: Causal Opacity and the Architecture of Honest AI Governance. DOI 10.5281/zenodo.19410730.