security: A one-prompt attack that breaks LLM safety alignment

The provided excerpt is only a short teaser and does not include the article’s technical details, findings, or mitigations.

• Source: Microsoft Security Blog