Mistral AI launches content moderation API to detect potentially harmful content in 11 languages, including Japanese
Mistral AI, a French artificial intelligence startup, has announced a moderation API for its large-scale language model (LLM)-based moderation system that supports multiple languages, including Japanese.
Mistral Moderation API | Mistral AI | Frontier AI in your hands
Moderation | Mistral AI Large Language Models
https://docs.mistral.ai/capabilities/guardrailing/
Mistral AI takes on OpenAI with new moderation API, negotiated harmful content in 11 languages | VentureBeat
https://venturebeat.com/ai/mistral-ai-takes-on-openai-with-new-moderation-api-tackling-harmful-content-in-11-languages/
The moderation API is powered by a fine-tuned version of Mistral AI's Ministral 8B and is designed to detect potentially harmful content across nine different categories, including sexual content, hate speech, violence, dangerous activities, and personally identifiable information.
Mistral AI's moderation API comes in two flavors, one for text and one for conversational content, and is natively multilingual, having been trained in English, Japanese, German, French, Italian, Portuguese, Spanish, Arabic, Chinese, Korean, and Russian.
According to Mistral AI, this moderation API is used in the moderation system of Mistral AI's chatbot service 'Le Chat'. Mistral AI said, 'Safety is important to make AI useful. We believe that system-level guardrails are essential to protect downstream deployments.'
VentureBeat, an IT news site, said, 'The release of Mistral AI's moderation API follows a series of recent partnerships by Mistral AI, including deals with Microsoft Azure, Qualcomm and SAP. Mistral AI has established itself as an increasingly important player in the enterprise AI market and announced that in October 2024 it will host Mistral models, including Mistral Large 2, on its own infrastructure, providing customers with a secure AI solution that complies with European regulations.'
VentureBeat also noted that 'Mistral AI's approach is particularly noteworthy because it focuses on both edge computing and comprehensive safety features,' and that Mistral AI's solution, which addresses data privacy and compliance concerns, will be attractive to European companies that are subject to strict data protection regulations.
Related Posts:
in Software, Posted by log1i_yk