01
Implement state-of-the-art AI ethics & security
We follow leading frameworks (OECD, UNESCO, NIST, open-source community standards), favor open models, optimize resources, mitigate bias, and adhere to the data scientist’s Hippocratic Oath.
AI MANIFESTO
Secure, ethical, responsible AI.
Our commitment to trust and sustainability. We build an agentic cyber governance platform, which makes our own AI discipline the first product we ship.
AI MANIFESTO · SEVEN ARTICLES
-
Implement state-of-the-art AI ethics & security
-
Ensure EU AI Act & regulatory compliance
-
Guarantee system security through rigorous testing
-
Reduce & monitor energy consumption
-
Ensure transparency in development & deployment
-
Protect user privacy & data
-
Monitor & continuously improve
Reviewed as the state of the art evolves. Reflected in how Ernest is engineered, operated and audited.
WHAT IS TRUSTWORTHY AI?
Trust is an engineering requirement, not a slogan.
Artificial intelligence opens immense opportunities for cyber governance, and real risks if deployed carelessly. We are committed to deploying AI systems that are responsible and worthy of trust: fair, proportionate in their use, secure and private by design.
For us, trustworthy AI rests on ethical principles implemented through “security and privacy by design”, transparency, explainability and resource efficiency, verified continuously, like everything else we do.
These seven commitments are public, binding, and reflected in how Ernest is engineered, operated and audited.
SEVEN COMMITMENTS
What we hold ourselves to.
Each commitment maps to concrete practices, reviewed as our systems and the state of the art evolve.
-
-
02
Ensure EU AI Act & regulatory compliance
Compliance-by-design, grounded in deep experience with the GDPR, NIS2, DORA and ISO 27001-family standards. As a European company, the EU AI Act is our baseline, not our ceiling.
-
03
Guarantee system security through rigorous testing
Models operated on secured platforms, continuous robustness evaluation during development and deployment, regular risk analyses and security reviews.
-
04
Reduce & monitor energy consumption
Environmental impact is tracked across the AI lifecycle, on reliable and verified infrastructure, with sobriety as an explicit evaluation criterion and KPIs to match.
-
05
Ensure transparency in development & deployment
Design and governance processes are documented; automated decisions and AI-generated content are clearly indicated; information is accessible to clients and regulators.
-
06
Protect user privacy & data
Carefully selected, consent-based datasets excluding unauthorized personal data; alignment with ANSSI and CNIL recommendations; privacy integrated at the design stage. Customer data never trains models.
-
07
Monitor & continuously improve
End-to-end AI risk management aligned with the NIST AI Risk Management Framework and ISO/IEC 42001: continuous monitoring, incident reporting, user feedback, ongoing adaptation.
Held to it, in public.
This manifesto evolves with the state of the art, and stays public so customers, regulators and candidates can hold us to it.