Meta has launched the Purple Llama project, which is related to its artificial intelligence tool of the same name.
In this case, the mentioned technology giant offers a kind of set of security solutions that help developers responsibly create their models of machine intelligence.
Meta introduced the Llama 2 artificial intelligence configuration in July. This project was implemented as part of the technology giant’s collaboration with Microsoft. This week, the company reported that Llama models have been downloaded more than 100 million times.
The technology giant announced that the components of its new solution will be licensed permissively. In this case, it will be possible to conduct research and carry out commercial use to promote cooperation and standardization of trust and security tools.
The company also said it would share what could be the industry’s first set of advanced cybersecurity assessments for large language models.
The new set of tools offered by the tech giant includes indicators for quantifying the risk of LLM cyber defense, ways to assess the frequency of unsafe code suggestions, and solutions that make it difficult for AI models to generate malicious code or its participate in cyber attacks.
Also this week, it became known that Meta is testing more than 20 functions of generative artificial intelligence. The technology giant carries out this activity on its virtual platforms, including Facebook, Instagram, Messenger, and WhatsApp. The new features relate to aspects of the functioning of social networks such as search, advertising, and business messaging.
