Topic: Security

Illustration of a protractor against a vibrant background split between blue and purple, surrounded by abstract black and pink patterns.
Published 
5 min read

Measurement is the key to helping keep AI on track  

Measurement is the key to helping keep AI on track  By Susanna Ray (() => { let $iframe, $closeBtn; const $player = document.getElementById(‘player-66fd9c5fe5a45’) ; const $playerPlayBtn = document.getElementById(‘player-btn-66fd9c5fe5a45’) ; const $playerPlayTitleBtn = document.getElementById(‘player-title-btn-66fd9c5fe5a45’) ; const $playerText = document.getElementById(‘player-text-66fd9c5fe5a45’) ; function onKeyDown(e) { if (document.activeElement !== $closeBtn) { $iframe.
Microsoft Source Microsoft Source
Illustration showing a red lock
Published 
5 min read

Red teams think like hackers to help keep AI safe  

Red teams think like hackers to help keep AI safe By Susanna Ray (() => { let $iframe, $closeBtn; const $player = document.getElementById(‘player-66fd9c030be4a’) ; const $playerPlayBtn = document.getElementById(‘player-btn-66fd9c030be4a’) ; const $playerPlayTitleBtn = document.getElementById(‘player-title-btn-66fd9c030be4a’) ; const $playerText = document.getElementById(‘player-text-66fd9c030be4a’) ; function onKeyDown(e) { if (document.activeElement !== $closeBtn) { $iframe.
Microsoft Source Microsoft Source
Photo of developer evaluating data from intelligent apps built in Azure
Published 
7 min read

AI jailbreaks: What they are and how they can be mitigated 

Microsoft security researchers, in partnership with other security experts, continue to proactively explore and discover new types of AI model and system vulnerabilities. In this post we are providing information about AI jailbreaks, a family of vulnerabilities that can occur when the defenses implemented to protect AI from producing harmful content fails.
Microsoft Security Blog Microsoft Security Blog
Graphic with illustrations of gift cards.
Published 
2 min read

Cyber Signals: Inside the growing risk of gift card fraud 

In the ever-evolving landscape of cyberthreats, staying ahead of malicious actors is a constant challenge. The latest edition of Cyber Signals dives deep into the world of Storm-0539, also known as Atlas Lion, shedding light on their sophisticated methods of gift and payment card theft.
Microsoft Security Blog Microsoft Security Blog