Anthropic’s Claude Mythos is still undergoing internal testing – but it is already becoming clear what its release could mean for IT security.

Claude Mythos, a new AI model from Anthropic, is currently only accessible to a closed group of test participants – companies involved in the internal security project Glasswing. A wider release is still pending, but is considered a possibility. What the preview version already reveals is cause for concern: the model is designed not only to detect unknown zero-day vulnerabilities in code, but also to exploit them autonomously and generate exploits – including N-day exploits from freshly released patches.

Should such a model become publicly available, IT departments would face massively increased pressure. Automated attacks, faster exploitation of vulnerabilities, targeted data manipulation – the question companies must ask themselves today is: Can I still trust my data at all?

BACKUP EAGLE® provides a clear response: complete data history, verifiable integrity and transparent reports – so that companies can prove that their data is accurate and unaltered at all times.