Microsoft's research reveals that a single prompt can unalign safety-trained AI models, demonstrating the fragility of pre-deployment safety measures in real-world applications.

Members of a Microsoft Corp. team tasked with using hacker tactics to find cybersecurity issues have open-sourced an internal tool, PyRIT, that can help developers find risks in their artificial ...

‘We can no longer talk about high-level principles,’ says Microsoft’s Ram Shankar Siva Kumar. ‘Show me tools. Show me frameworks.’ Generative artificial intelligence systems carry threats new and old ...

In the realm of IT security, the practice known as red teaming -- where a company's security personnel play the attacker to test system defenses -- has always been a challenging and resource-intensive ...

Red teaming AI systems is a complex, multistep process. Microsoft’s AI Red Team leverages a dedicated interdisciplinary group of security, adversarial machine learning, and responsible AI experts. The ...

Microsoft's research reveals that a single prompt can dismantle safety training in AI models, highlighting the fragility of ...

Microsoft has open sourced a key piece of its AI security, offering a toolkit that links data sets to targets and scores results, in the cloud or with small language models. At the heart of ...

I wore the world's first HDR10 smart glasses TCL's new E Ink tablet beats the Remarkable and Kindle Anker's new charger is one of the most unique I've ever seen Best laptop cooling pads Best flip ...

Microsoft says it needs to strengthen trust in AI, which cannot be achieved without a full understanding of the changes AI generates. In Microsoft’s AI Red Team, alongside cybersecurity experts, ...

Many risk-averse IT leaders view Microsoft 365 Copilot as a double-edged sword. CISOs and CIOs see enterprise GenAI as a powerful productivity tool. After all, its summarization, creation and coding ...