|
|
We anticipate the attacker's next move, and we explain why.
A predictive cybersecurity module built to move security operations from reactive detection to anticipatory resilience.
AuraSec has received funding from the European Union's Horizon Europe research and innovation programme through the Resilmesh Open Call #2 issued and executed under the Resilmesh project (Grant Agreement no. 101119681).
The challenge
Security operations are extraordinarily good at reporting what has just happened, and far less good at anticipating what happens next. In regulated sectors that lag carries a material cost: by the time the picture is complete, the options for containing the incident have already narrowed.
What AuraSec does
Instead of an abstract risk score, AuraSec is designed to name the specific move an attacker is likely to make next, with a time window and a probability.
Every prediction is designed to arrive with its reasoning. An unexplained prediction is not actionable: an analyst cannot act on a black box, and an auditor cannot accept one.
If the system drifts outside the conditions it was built for, it steps aside and conventional rules resume control. Security degrades to standard, never to none.
Open standards and documented interfaces throughout. The ResilMesh connector is released as open source under Apache 2.0, so the integration remains verifiable and reusable by third parties.
Each of these capabilities rests on a detailed technical design. Request a technical briefing.
Targets and open data
These are the targets the module is built to meet, and the criteria against which each phase is validated. Measured results are published as the validation progresses.
≤ 250 ms
Inference latency at the edge, in near real time
≥ 80 %
Accuracy predicting the next MITRE ATT&CK technique, in lab conditions
< 5 %
False positive rate
100 %
Explainability coverage: every alert carries its feature attribution, its causal trace and a confidence score
≥ 40 %
Improvement in detection time
≥ 30 %
Accuracy gain on silent and multi-vector threats
The evaluation phase produces an irreversibly anonymised adversarial dataset, annotated against MITRE ATT&CK and the ResilMesh zone taxonomy, built to FAIR principles and published on the project's Zenodo instance. It is not available yet: it is released once the final validation closes.
The ResilMesh connector, the API specifications, the metadata schemas and the anonymised datasets are released as open source under Apache 2.0, with a STIX 2.1 compliant, engine-agnostic design. The predictive engine and the explainability module are distributed under a commercial licence, with access rights granted to the ResilMesh partners and the European Commission for evaluation, demonstration and research replication.
The project
AuraSec is being developed and validated as a module of the ResilMesh platform, taking the module from TRL 4 to TRL 7 in nine months, with the final phase running in a European financial-sector environment.
Architecture, data governance and the plan against which the module would be measured.
Integration into the platform, performance work, and the explainability layer in the analyst's dashboard.
Operational validation in a real-world environment and a public demonstration at the ResilMesh final event.
Who is building it
AuraSec is developed by Tesla Technologies & Software S.L., an innovation-driven Spanish company headquartered in Santiago de Compostela, with hubs in Madrid, Manchester, Arad and Buenos Aires and more than fourteen years delivering AI and digital-engineering solutions.
The sub-project runs under ResilMesh Open Call #2, part of the EU's Horizon Europe programme, with technical mentoring from the ResilMesh consortium.
About ResilMesh
ResilMesh is a Horizon Europe project building cyber situational awareness for distributed infrastructures: it collects, normalises and correlates security telemetry across zones so that operators can see and reason about an incident as one picture rather than as scattered alerts.
AuraSec sits inside that platform as a module, and reached it through the project's second open call for third-party sub-projects. The consortium publishes its own results, newsletters and open calls on its channels.
We are available to security teams, platform vendors and research groups for technical discussion, whether on how the module works, on integration, or on collaboration.
Tesla Technologies & Software S.L. · Santiago de Compostela, Spain