Microsoft releases inaugural AI transparency report

In its Responsible AI Transparency Report, primarily covering 2023, Microsoft touts its achievements in safely deploying AI products. The annual AI transparency report is one of the commitments the company made after signing a voluntary agreement with the White House in July of last year. Microsoft and other companies promised to establish responsible AI systems and commit to security.

Microsoft says in the report that it created 30 responsible AI tools last year, grew its responsible AI team, and required teams to build generative AI applications to measure and map risks throughout the development cycle. The company notes that it added Content Credentials to its imaging platforms, which places a watermark on a photo and labels it as created by an AI model.

The company says it has given Azure AI customers access to tools that detect problematic content such as hate speech, sexual content, and self-harm, as well as tools to assess security risks. This includes new jailbreak detection methods, which were expanded in March this year to include indirect injections where malicious instructions are part of the data ingested by the AI ​​model.

It is also expanding its Red Team efforts, including both internal red teams deliberately trying to bypass security features in their AI models and red teaming applications to enable third-party testing before releasing new models.

However, your red team units have their work cut out for them. The company’s AI launches have not been immune to controversies.

Natasha Crampton, director responsible for AI at Microsoft, says in an email sent to The edge that the company understands that AI is still a work in progress and so is responsible AI.

“Responsible AI has no finish line, so we will never consider our work done under voluntary AI commitments. But we have made great progress since we signed them and we hope to build on our momentum this year,” says Crampton.