On 9 November 2026, we temporarily shut down our platform, after we observed suspicious activity that turned out to be a cyberattack. We restored our products only after we had removed the attacker from our systems and put extensive additional security measures in place. All services are now up and running again and no signs of malicious activity remain. We understand that this incident nevertheless causes concern for our customers. It was therefore our highest priority to conduct a thorough assessment of our systems and the incident, and we are sharing all findings here in full transparency.
What happened?
On the afternoon of 9 November 2026, we detected suspicious activity on our platform, which we identified as an attack by an autonomous, agent-based system. As a precaution, we temporarily shut down our platform. On the procedural side, we immediately began containment: We revoked the attacker’s access, replaced all affected credentials, closed the vulnerabilities exploited, and then brought our systems back online step by step over the following days, each only after it was confirmed secure. The last successful unauthorized action took place on the morning of 10 November 2026, and every attempt since then has been blocked.
Our engineering team has since reconstructed the attacker’s full path through our systems based on the logs of all affected services. We have filed a criminal complaint with the police and notified the data protection authority. Who is behind the attack remains unknown.
The following timeline records our findings and response actions (all times CET):
How could this happen?
Access was obtained via a combination of issues with an external database provider. Combined information contained in the source code of an outdated version of an open source tool built by Lyceum, together with a security vulnerability at our external database provider, provided the attacker with an entrypoint into a part of our internal operation systems. We disclosed this security vulnerability with the external database provider, who are releasing a patch, and removed all old versions of the open source tool.
Was customer data affected?
While on our platform, the attacker had access to our user database, so we consider it compromised. This affects all users who registered on our platform and their company registration data, including email addresses, contact names, company addresses, and phone numbers, if one was submitted upon registration. We are therefore treating this as a personal data breach under GDPR and have notified the Berlin data protection authority. Payment card, bank details and Lyceum API keys were not affected.
Furthermore, in line with our zero data retention principle, no prompts or outputs were stored that could have been exposed.
What are we doing to prevent this from happening again?
We set out a plan and started implementing significant changes to harden our infrastructure in the areas the attacker used, such as, but not limited to:
- Contact Credentials: Regular rotation of all production credentials and automated checks of every release before publication.
- Isolation: Further, full separation of customer workloads from our internal network.
- Authentication: A login requirement for every internal tool and service.
- Monitoring: Full request logging and automated alerts for unusual activity.
- Red teaming: Continuous penetration testing with uncensored and security models of our own infrastructure.
If you have any questions or concerns, please email us at [email protected].