The translation of the provided data: Swiss startup Lakera has recently launched an API designed to protect businesses from malicious prompt attacks by large language models. They have developed an interactive game called "Gandalf" to establish an attack classification system and are dedicated to safeguarding data security and filtering inappropriate content for children. Lakera's API offers businesses a comprehensive solution to counter the threats of malicious attacks from LLMs.