Architecture-aligned attack

An attack that works by abusing the intended design of a platform rather than breaking a technical control. In browser security, that means using legitimate permissions, update channels, and runtime features to perform surveillance, exfiltration, or session abuse while staying inside expected behavior.

Expanded Definition

An architecture-aligned attack succeeds by staying within the product’s intended trust boundaries, feature set, and permission model instead of bypassing them. In browser security, that can mean using legitimate extension permissions, normal update mechanisms, service workers, clipboard access, or session handling to observe activity, move data, or persist control. The attacker is not “breaking” the platform so much as turning its design assumptions into an advantage.

This makes the term different from exploit-centric language. A software vulnerability often implies a flaw that must be patched. An architecture-aligned attack may not involve a defect at all, only an implementation that grants too much capability once a benign feature is installed or approved. Guidance in MITRE ATT&CK Enterprise Matrix helps analysts map the resulting behaviours, while the browser and AI security community increasingly uses the term to describe abuse of legitimate runtime paths rather than code injection. Definitions vary across vendors, and no single standard governs this yet.

The most common misapplication is calling any stealthy intrusion architecture-aligned, which occurs when the attacker actually relies on a broken control rather than on intended platform behaviour.

Examples and Use Cases

Implementing browser or agent security rigorously often introduces friction, because tighter permission review and runtime isolation can reduce convenience for users and automation teams while lowering the attack surface.

• A malicious browser extension requests broad but apparently legitimate access, then reads page content, session artefacts, or form inputs without exploiting a software bug.
• An attacker abuses the normal extension update path to deliver new surveillance code after the extension has already been approved, turning a trusted channel into a persistence mechanism.
• A workspace or identity tool exposes valid automation hooks, and a compromised account uses those hooks to export data or trigger actions that look like ordinary user workflow.
• An AI browser agent is instructed through normal prompts and tool calls to retrieve sensitive content, making the abuse resemble intended orchestration rather than malware execution; Anthropic’s first AI-orchestrated cyber espionage campaign report shows why legitimate tooling can be repurposed for covert operations.
• Threat hunters correlate suspicious but policy-compliant behaviour with campaign patterns in MITRE ATT&CK Enterprise Matrix and support it with CISA cyber threat advisories when the same tradecraft appears across multiple incidents.

Why It Matters for Security Teams

Architecture-aligned attacks matter because defensive programs often focus on patching vulnerabilities while overlooking abuse of trusted design paths. That gap is especially relevant in browser ecosystems, SaaS platforms, identity workflows, and agentic AI systems where permissions, tokens, and tool access are intentionally powerful. If those capabilities are granted too widely, an attacker can exfiltrate data or manipulate sessions without triggering classic exploit detection.

For identity and access teams, the lesson is that authorization boundaries must be treated as security controls, not just user experience choices. This aligns naturally with control thinking in NIST SP 800-53 Rev 5 Security and Privacy Controls, especially where least privilege, access monitoring, and auditability are required. The concept also intersects with MITRE ATLAS adversarial AI threat matrix when AI agents are manipulated through their approved interfaces rather than through model compromise.

Organisations typically encounter the operational impact only after an apparently legitimate feature has been abused at scale, at which point architecture-aligned attack analysis becomes operationally unavoidable to contain the abuse.

Related resources from NHI Mgmt Group

• Why do AI-enabled attack chains change the value of Zero Trust Architecture?
• Why does Agentic AI make NHI attack surface expand so significantly?
• How does NHI security relate to Zero Trust Architecture?
• Why do non-human identities complicate zero trust architecture?