Targeted ISMS hardening against Frontier AI attacks. MRIS assesses which security controls still work against autonomously operating AI attackers – and which only work on paper.
MRIS assesses established security standards control by control against GenAI-accelerated attacks – currently ISO/IEC 27001 (Annex A) and TISAX® 2027 –, points to concrete Mythos-Hardening Controls (MHC) from a cross-standard catalog and operationalises them with MRIS.bd.1 as a user-defined module for BSI IT-Grundschutz.
Mythos (Anthropic) was the first frontier AI model whose capabilities in an attacker’s hands fundamentally changed the security landscape. MRIS uses it as the reference threat model: every control is measured against an attacker with Mythos-level capabilities. In public discourse, systems of this capability class are called Frontier AI or agentic AI – within MRIS, "Mythos" is the vendor-neutral name for exactly this class.
Implement once, benefit repeatedly: each MHC is defined across standards – the same measure closes gaps in every framework that references it and yields audit evidence for several assessments at once. Instead of maintaining a separate catalog per framework, you prioritize one central catalog by breadth of impact. It grows with every standard assessed.
Protects long-lived confidential data captured today from later decryption (“harvest now, decrypt later”) and, with the cryptographic inventory, lays the groundwork for an orderly migration to post-quantum algorithms.
Breadth of impact: 1 ISO controls · 1 IT-GS requirement
Makes the origin and components of every piece of software verifiable – the basis for detecting AI-accelerated supply-chain attacks and tampered builds before they reach production.
Breadth of impact: 6 ISO controls · 1 IT-GS requirement
Renders intercepted or real-time-relayed credentials worthless, removing the basis for automated, AI-scaled phishing.
Breadth of impact: 5 ISO controls · 10 TISAX® findings · 1 IT-GS requirement
Ties access to verified workload identities instead of network zones – reducing lateral movement once an AI attacker has initial access, and enabling the safe use of AI agents.
Breadth of impact: 7 ISO controls · 1 IT-GS requirement
Detects unknown, signature-less malware by its behavior as well as attacks deliberately split into many small, individually inconspicuous steps – in their sum rather than step by step.
Breadth of impact: 11 ISO controls · 9 TISAX® findings · 1 IT-GS requirement
Hardens runtime environments through signed images and attestation – a compromised artifact does not reach production unnoticed, and data stays protected even during processing.
Breadth of impact: 2 ISO controls · 1 IT-GS requirement
Enforces demonstrable separation between tenants – a compromised tenant does not become a springboard to others; effectiveness is proven by separation tests, not merely asserted.
Breadth of impact: 2 ISO controls · 1 IT-GS requirement
Ensures backups remain restorable even with compromised admin rights, and that recovery actually works in an emergency instead of existing only on paper.
Breadth of impact: 5 ISO controls · 3 TISAX® findings · 1 IT-GS requirement
Shortens the time between a known vulnerability and its remediation to a pace that keeps up with AI-accelerated exploit development.
Breadth of impact: 6 ISO controls · 6 TISAX® findings · 1 IT-GS requirement
Closes the window between two periodic review cycles in which new attack patterns, incidents or system changes would otherwise remain undetected until the next round.
Breadth of impact: 4 ISO controls · 18 TISAX® findings · 1 IT-GS requirement
Shortens the response time to a security-relevant event to a level that no longer depends on human processing speed alone.
Breadth of impact: 4 ISO controls · 4 TISAX® findings · 1 IT-GS requirement
Validates the entire defense chain against realistic, AI-driven attack scenarios – revealing whether detection and response work in practice rather than only on paper.
Breadth of impact: 2 ISO controls · 1 IT-GS requirement
Prevents an AI-powered assistant from independently pulling in an unvetted external tool, service or software package while working on a task, and additionally constrains the identity, prompt-injection exposure and response time of agents in production.
Breadth of impact: 3 ISO controls · 4 TISAX® findings · 1 IT-GS requirement
Prevents an AI-assisted, convincingly forged or embellished supplier attestation from being accepted as sufficient without detection.
Breadth of impact: 5 TISAX® findings
Prevents a convincing deepfake identity or forged document from being accepted undetected in hiring or facility-access processes.
Breadth of impact: 6 TISAX® findings
Limits how far a manipulated, faulty or compromised agent can act within a chain or orchestration of multiple AI agents before damage occurs or spreads.
Breadth of impact: TISAX®: original
Catalog status: 16 MHC · August 2026. * MHC-16 is an original addition beyond the TISAX® diagnosis. Seven MHC (highlighted) apply across all three frameworks; MHC-01 to MHC-13 are additionally operationalised as IT-Grundschutz requirements (MRIS.bd.1) – one implementation, effect across up to three frameworks.
93 controls assessed individually: 29 robust, 37 partially degraded, 4 pure friction, 23 not affected. 13 MHC close the gaps.
View the full assessment → TISAX® ISA 2027 · Information Security45 controls (324 requirements) assessed individually: 4 robust, 26 partially degraded, 1 pure friction, 14 not affected. 72 findings, carried by 10 MHC.
View the full assessment → BSI IT-Grundschutz · MRIS.bd.1 v1.214 requirements operationalise the 13 MHC as a user-defined module – with implementation guidance, YAML and cross-reference table. Not approved by the BSI.
View the module →License: CC BY-NC 4.0 · © 2026 Richard Peddi · No registration required.
The MRIS publications are working aids for assessing the effectiveness of existing security controls. They assume an established ISMS and do not replace it. They do not constitute legal, compliance or certification advice, make no claim to completeness and establish no warranty. Use is at your own responsibility.
TISAX® is a registered trademark of the ENX Association; the ENX Association administers TISAX® on behalf of the German Association of the Automotive Industry (VDA), which develops and publishes the underlying requirements catalog (VDA ISA). MRIS is an independent, privately produced work with no affiliation to the ENX Association or the VDA. It is neither part of the official TISAX® assessment process nor authorized, reviewed or endorsed by ENX, and does not replace a TISAX® maturity assessment by an accredited audit provider.