Leading cloud providers, AI model developers and cybersecurity agencies have publicly warned that AI has lowered the cost and raised the scale and credibility of attacks: targeted phishing, voice and video impersonation, accelerated guessing of weak passwords and intrusions orchestrated by automated agents. Two technical clarifications worth stating plainly:
- AI does not break modern encryption; it exploits the human factor. Defeating AES-256 and the schemes on this page does not depend on the attacker's intelligence but on mathematical capabilities that do not exist today. Real AI-assisted attacks enter through a convincing email, a falsified video call or a reused password. That is why our defenses prioritize phishing-resistant identity, second-channel verification and anomaly detection.
- The state of the art evolves, and so do we. We follow NIST standards, including the already-published post-quantum cryptography, and will migrate components as the industry adopts them. Our contractual commitment is to keep measures at the state of the art and update them at the pace of the threat, not the calendar.
Second-channel protocol. No instruction received by email, message or phone call that seeks to change credentials, bank accounts or information destinations is executed without confirming it with the authorized person through a previously agreed second channel, even if it appears to come from you or from us. We ask our customers for the same care in the opposite direction.