Mitigant and Cyberlift: Testing a Reinsurer’s AWS Cloud Under Real Attack
How do you know your cloud is secure before you move your most sensitive systems into it? You attack it first. That was the thinking behind a new customer story from ECA member Mitigant and Paris-based cloud security consultancy Cyberlift. The field report was published in September 2026.
The challenge
A French reinsurer planned to migrate sensitive systems to Amazon Web Services (AWS). Before doing so, it wanted to know how its cloud environment would hold up against real-world attacks. It had three goals:
- Test the environment against realistic attack scenarios.
- Measure how well its teams detect and respond under real conditions.
- Get an action plan prioritised by risk.
A Franco-German partnership
The project combined Mitigant’s cloud attack emulation platform with Cyberlift’s expertise. Mitigant provides more than 40 ready-to-use attacks across over 15 AWS services, run automatically and mapped to the MITRE ATT&CK framework. Cyberlift scoped the scenarios around the reinsurer’s business priorities and ran them safely in a sensitive environment. It then turned the results into a clear, risk-based action plan for both executives and technical teams.
One scenario shows how real the tests were. It simulated an attacker who switches off logging, copies a storage bucket to an attacker-owned account and encrypts the data with their own key. That combination of silent data theft and ransomware makes recovery impossible without a backup.
The results
65% of the simulated attacks succeeded. The exercise produced more than 40 recommendations and over 10 best practices. They fall into five areas: blocking attacks, fixing issues automatically, alerting, auditing and strengthening the security baseline. The reinsurer now has evidence of which attack paths can actually be exploited, fewer false positives, and a four-phase plan to close the gaps before migration.
The lesson, in the partners’ words: “security controls only prove their worth when tested against realistic Cloud-native attacks.”
ATOS
With Atos Genesis Intelligence and its Atos Codex component, the French digital services firm aims to link models, data and business operations to automate decision-making in sensitive environments. This new offering comes as Atos gradually emerges from its financial crisis, though it remains far from returning to growth.
Atos aims to go beyond text generation and recommendations. The group is launching Atos Genesis Intelligence, a suite combining data management, model deployment, agent orchestration and governance. Its centrepiece, Atos Codex, ‘Centre of Operations and Data EXcellence’, is designed to convert information produced or analysed by AI into operational actions in critical environments.
The suite brings together the building blocks needed to scale up use cases: open architecture; deployment in a public or sovereign cloud, on-premises or in an isolated environment; a data foundation with quality and traceability features; selection of the appropriate model for each query to control costs; a ‘factory’ for small language models (SLMs) that can be deployed locally; and a portal for customisable AI agents.
Read more here
