Claude Mythos 5.1: How Codafication Is Strengthening Software Security Against Evolving AI Cyber Threats

“Are you Mythos ready?”

AI cyber capabilities have become a global concern, with exploitation routinely happening close to, or before disclosure. Frontier models are lowering the barrier to sophisticated attacks: they consistently perform well in industry evaluations and have been utilised at least one state-backed campaign.

These same capabilities, in the hands of defenders, can be used to help mitigate these developing risks and add an additional layer to security testing.

Codafication has been accepted into Anthropic’s Cyber Verification Program at the Red Team Access tier and is actively using Claude Mythos 5.1 to test its own software and strengthen security.

What Is Claude Mythos 5.1?

Claude Mythos 5.1 is Anthropic’s latest Mythos AI model, with advanced capabilities in software engineering, complex reasoning and cybersecurity.

Originally introduced through Mythos Preview in April 2026, the model family demonstrated an ability to identify previously unknown software vulnerabilities and develop working exploits, with a particular strength in attack-chaining low-level vulnerabilities into serious exploits.

The same technology that could accelerate malicious attacks can also help security professionals identify weaknesses, investigate vulnerabilities and strengthen software defences.

Project Glasswing and the Cyber Verification Program

In April 2026, Anthropic launched Project Glasswing, bringing together leading technology and infrastructure organisations to identify and address vulnerabilities in critical software.

The initiative’s 12 launch partners included: Amazon Web Services (AWS), Anthropic, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, Linux Foundation, Microsoft, NVIDIA & Palo Alto Networks.

The initiative demonstrated the potential of AI-assisted security testing. Anthropic reported that Project Glasswing partners had identified at least 129,000 verified software vulnerabilities between April and July.

In October 2026, Anthropic integrated Project Glasswing into its expanded Cyber Verification Program, providing qualifying security professionals with tiered access to advanced AI cybersecurity capabilities, including Mythos 5.1.

Codafication’s Red Team Access: Expanded Security Testing

Codafication has been accepted into the Cyber Verification Program’s Red Team Access tier, providing access to advanced cybersecurity capabilities and a broader scope of authorised testing activities restricted from public access.

Beyond standard defensive vulnerability analysis, Red Team Access supports authorised penetration testing and adversarial security tasks, with fewer restrictions on legitimate cybersecurity activities.

This enables security teams to investigate a wider range of potential attack scenarios and assess software from an adversarial perspective, within approved testing boundaries.

Codafication is actively using Claude Mythos 5.1 to assess its own software, investigate potential vulnerabilities and inform ongoing security improvements.

As an insurance technology provider, our platforms, including Crunchwork and Unity Cloud, support interconnected systems, stakeholders and operational workflows where security is essential.

By incorporating these capabilities into our security practices, we are taking a proactive approach to understanding and addressing emerging AI cybersecurity risks.

What Does Mythos Readiness Mean for Technology Suppliers?

For organisations evaluating technology suppliers, Mythos readiness goes beyond awareness of emerging AI threats. It involves taking practical steps to assess vulnerabilities, strengthen defences and maintain established security controls.

Codafication’s use of Mythos 5.1 complements our existing Information Security Management System and security tooling across our Crunchwork, Unity Cloud and Virtual products. It joins a large list of existing practices including our independent external penetration testing, ISO 27001, and SOC2 Type II auditing.

Strengthening Security as AI Evolves

As AI continues to reshape cybersecurity, organisations must evolve their defensive capabilities alongside emerging threats.

Codafication’s Red Team Access and active use of Claude Mythos 5.1 reflect our commitment to staying ahead of these developments through practical, responsible security testing.

Being Mythos ready isn’t about claiming immunity from cyber threats. It’s about actively using emerging technology to identify risks, strengthen software and continuously improve security.