12:57 PM EDT, 03/28/2024 (MT Newswires) -- Microsoft ( MSFT ) reported Thursday a number of new tools to its Azure AI Studio to detect and block so-called prompt injection attacks that manipulate an AI system into doing something outside its intended purpose.

The company said it would be launching Prompt Shields to prevent such attacks and also introduce groundedness detection to detect "hallucinations" in model outputs.

Microsoft ( MSFT ) also said Azure AI Studio will also begin supporting safety evaluations to assess an application's vulnerability to jailbreak attacks and to generating content risks.

Price: 419.79, Change: -1.64, Percent Change: -0.39