Microsoft’s Yonatan Zunger on Red Teaming Generative AI

In this episode of the Microsoft Threat Intelligence Podcast host Sherrod DeGrippo is joined by Christian Dameff and Jeff Tully, co-directors from the UCSD Center for Healthcare Cybersecurity, and contributors to our recent Healthcare Ransomware report.   They discuss their unique backgrounds as doctors and hackers, focusing on healthcare cybersecurity, and the growing risks of hospital ransomware attacks. Christian shares his journey from hacking as a teenager to combining his passion for medicine and cybersecurity, particularly the risks posed to patient safety by vulnerable medical devices. Jeff adds his perspective, highlighting the parallels between medicine and hacking, and their efforts at UCSD to bring evidence-based research to healthcare cybersecurity. The conversation explores the challenges and importance of protecting critical healthcare systems from cyber threats, aiming to improve patient safety and outcomes.  In this episode you’ll learn:       How medical device vulnerabilities reveal the impact of cybersecurity on patient care  The lack of comprehensive data on healthcare ransomware attacks  When ransomware-induced disruptions can delay life-saving procedures    Some questions we ask:         As healthcare providers, what stands out to you about ransomware in healthcare?  What does the UCSD Center for Healthcare Cybersecurity do?  What ransomware attacks are common in healthcare, and how do they differ from other industries?    Resources:   View Jeff Tully on LinkedIn   View Christian Dameff on LinkedIn   View Sherrod DeGrippo on LinkedIn   Healthcare Ransomware Report    Related Microsoft Podcasts:                    Afternoon Cyber Tea with Ann Johnson  The BlueHat Podcast  Uncovering Hidden Risks        Discover and follow other Microsoft podcasts at microsoft.com/podcasts     Get the latest threat intelligence insights and guidance at Microsoft Security Insider      The Microsoft Threat Intelligence Podcast is produced by Microsoft and distributed as part of N2K media network.

In this episode of the Microsoft Threat Intelligence Podcast host Sherrod DeGrippo is joined by Microsoft’s Dinesh Natarajan, Senior Threat Hunter, and Thomas Ball, Senior Security Researcher. They unpack recent findings around AsyncRAT, a remote access Trojan (RAT) used for keylogging, data exfiltration, and deploying further malware.     Dinesh explains how attackers are now using screen-sharing tools, like Screen Connect, as part of a new infection chain that makes the malware delivery process more deceptive. Thomas then shares insights on SectopRAT, another threat targeting browser data and crypto wallets. Uniquely, this RAT creates a second desktop, allowing attackers to operate undetected.     Next, Sherrod talks with Microsoft’s Senior Director of Diplomacy, Kaja Ciglic, about the UN’s proposed cybercrime treaty. Originally spearheaded by Russia, the treaty aims to create a global framework for prosecuting cybercrime, but critics worry about its potential impact on freedom of expression and human rights.    In this episode you’ll learn:       How tech support scam emails lead to AsyncRAT installations on different devices  The importance of leveraging tools like Microsoft Defender's SmartScreen for protection  The treaty encourages cooperation but may let governments exploit unclear cybercrime definitions    Some questions we ask:     How does social engineering through email play a role in these attacks?  What capabilities does AsyncRat have, and why is it so concerning?  How do we ensure the treaty doesn't impact freedom of expression or human rights?    Resources:   View Dinesh Natarajan on LinkedIn  View Thomas Ball on LinkedIn  View Kaja Ciglic on LinkedIn   View Sherrod DeGrippo on LinkedIn     Related Microsoft Podcasts:                    Afternoon Cyber Tea with Ann Johnson  The BlueHat Podcast  Uncovering Hidden Risks        Discover and follow other Microsoft podcasts at microsoft.com/podcasts     Get the latest threat intelligence insights and guidance at Microsoft Security Insider      The Microsoft Threat Intelligence Podcast is produced by Microsoft and distributed as part of N2K media network.

In this episode of the Microsoft Threat Intelligence Podcast host Sherrod DeGrippo is joined by Proofpoint’s Greg Lesnewich and Microsoft’s Greg Schloemer to share the unique threat posed by North Korea’s (DPRK) state-sponsored cyber activities. The Gregs discuss their years of experience tracking North Korean cyber actors and the distinct tactics that set DPRK apart from other nation-sponsored threats. The conversation also explores North Korea’s high stakes, as DPRK threat actors operate under intense pressure from government handlers, adding a layer of urgency and fear to their operations. They share insights into North Korea’s aggressive use of stolen cryptocurrency to fund the regime’s initiatives, like ballistic missile tests, and discuss the broader geopolitical impact.   In this episode you’ll learn:       The technical sophistication and the relentlessness of DPRK cyber tactics  Complex motives behind funding and sustaining the North Korean government  The training and skills development of North Korean cyber operators    Some questions we ask:      How do North Korean threat actors set up their relay networks differently?  What sets North Korea apart from other nation-sponsored threat actors?  How do North Korean cyber actors differ from traditional e-crime actors?    Resources:   View Greg Schloemer on LinkedIn   View Greg Lesnewich on LinkedIn  View Sherrod DeGrippo on LinkedIn     Blog links:  Citrine Sleet Observed Exploiting Zero Day  New North Korean Threat Actor Identified as Moonstone Sleet  East Asia Threat Actor Technique Report    Related Microsoft Podcasts:                    Afternoon Cyber Tea with Ann Johnson  The BlueHat Podcast  Uncovering Hidden Risks        Discover and follow other Microsoft podcasts at microsoft.com/podcasts     Get the latest threat intelligence insights and guidance at Microsoft Security Insider      The Microsoft Threat Intelligence Podcast is produced by Microsoft and distributed as part of N2K media network.

In this episode of the Microsoft Threat Intelligence Podcast host Sherrod DeGrippo is joined by Yonatan Zunger, CVP of AI Safety and Security at Microsoft. The conversation delves into the critical role of the AI Red Team, which focuses on identifying vulnerabilities in AI systems. Yonatan emphasizes the importance of ensuring the safety of Microsoft’s AI products and the innovative methods the team employs to simulate potential threats, including how they assess risk and develop effective responses. This engaging dialogue offers insights into the intersection of technology, security, and human behavior in the evolving landscape of AI.     In this episode you’ll learn:          Why securing AI systems requires understanding their unique psychology  The importance of training and technical mitigations to enhance AI safety  How financial incentives drive performance improvements in AI systems  Some questions we ask:         How does Retrieval Augmented Generation (RAG) work?  What are the potential risks with data access and permissions in AI systems?  Should users tell language models that accuracy affects their rewards to improve responses?  Resources:   View Yonatan Zunger on LinkedIn   View Sherrod DeGrippo on LinkedIn     Related Microsoft Podcasts:                    Afternoon Cyber Tea with Ann Johnson  The BlueHat Podcast  Uncovering Hidden Risks      Discover and follow other Microsoft podcasts at microsoft.com/podcasts     Get the latest threat intelligence insights and guidance at Microsoft Security Insider      The Microsoft Threat Intelligence Podcast is produced by Microsoft and distributed as part of N2K media network.

In this episode of the Microsoft Threat Intelligence Podcast, host Sherrod DeGrippo is joined by security researchers Anna and Keivan to discuss two prominent threat actors: Vanilla Tempest and Peach Sandstorm.  Vanilla Tempest, a financially motivated cybercrime group, has been involved in recent ransomware attacks on U.S. hospitals, utilizing various ransomware payloads such as Ink. They are known for using tools like PowerShell scripts and Goot Loader to exfiltrate data and extort victims. Peach Sandstorm, an Iranian nation-state threat actor, focuses on cyber espionage and intelligence collection. They have targeted various sectors, including energy, defense, and critical infrastructure, and have shown increasing sophistication in their attacks. Later, Sherrod speaks with Colton Bremer, a senior security researcher at Microsoft, about his work on the Defender Experts (DEX) team. Colton explains the different tiers of DEX services, which focus on detecting and mitigating advanced threats that may bypass traditional security measures.  In this episode you’ll learn:       A backdoor called Tickler that uses Azure infrastructure for command and control  The significance of these groups' tactics and maintaining ransomware resiliency  The different tiers of DEX services detecting and mitigating advanced threats    Some questions we ask:     How does Vanilla Tempest typically execute their attacks?  Has Peach Sandstorm evolved over time in their cyber espionage efforts?  What can individuals or organizations do to mitigate cloud identity abuse?    Resources:   View Colton Bremer on LinkedIn   View Sherrod DeGrippo on LinkedIn     Related Microsoft Podcasts:                    Afternoon Cyber Tea with Ann Johnson  The BlueHat Podcast  Uncovering Hidden Risks        Discover and follow other Microsoft podcasts at microsoft.com/podcasts     Get the latest threat intelligence insights and guidance at Microsoft Security Insider      The Microsoft Threat Intelligence Podcast is produced by Microsoft and distributed as part of N2K media network.

In this episode of the Microsoft Threat Intelligence Podcast host Sherrod DeGrippo is joined by Nick Monaco, Principal Threat Intelligence Analyst at Microsoft, delving into findings from Microsoft's April 2024 East Asia threat report. They discuss Gingham Typhoon's expanding cyber operations in the South Pacific, notably targeting strategic partners like Papua New Guinea despite their involvement in China's Belt and Road Initiative. The conversation shifts to Nylon Typhoon's global espionage efforts, including recent activities in South America and Europe. They also cover Volt Typhoon's sophisticated attacks on U.S. critical infrastructure and highlight Storm 1376's (now Tides of Flood) use of AI-generated news anchors for spreading misinformation. This episode emphasizes the evolving nature of cyber threats and influence operations, including the creative use of technology by adversaries to advance their agendas.  * This episode is from April 2024 and is not new information.    In this episode you’ll learn:          How Nylon Typhoon targets geopolitical intelligence in South America and Europe  The evolving landscape of influence operations and China's growing capabilities  How disinformation campaigns have exploited real-world events    Some questions we ask:         How has generative AI changed influence operations and disinformation?  What are the key trends in North Korean cyber operations with cryptocurrency and AI?  Why are Chinese influence operations engaging with questions on social media?    Resources:   View Nick Monaco on LinkedIn   View Sherrod DeGrippo on LinkedIn     Related Microsoft Podcasts:                    Afternoon Cyber Tea with Ann Johnson  The BlueHat Podcast  Uncovering Hidden Risks      Discover and follow other Microsoft podcasts at microsoft.com/podcasts     Get the latest threat intelligence insights and guidance at Microsoft Security Insider  The Microsoft Threat Intelligence Podcast is produced by Microsoft and distributed as part of N2K media network.

In this episode of the Microsoft Threat Intelligence Podcast host Sherrod DeGrippo is joined by the authors of the new book The Definitive Guide to KQL: Using Kusto Query Language for Operations, Defending, and Threat Hunting. Guests Rod Trent, Matt Zorich, and Mark Morowczynski discuss the significance of KQL (Kusto Query Language) in cloud data security and how it enables efficient data querying for threat detection in Microsoft products like Sentinel and Defender. They share insights from their own experiences, highlight key features of the book, and explain how both beginners and experts can benefit from KQL. Later in the episode Sherrod speaks with Senior Threat Hunter Lekshmi Vijayan about the growing trend of cyberattacks using malicious PowerShell commands. Lekshmi explains how attackers trick users into copying and pasting harmful code, often through compromised websites or phishing emails. They discuss how these attacks aim to install remote access tools like NetSupport RAT or information stealers, targeting sensitive data like browser credentials and crypto keys.    In this episode you’ll learn:       How KQL is applied in real-world security scenarios including incident response  Key features and benefits of KQL when it comes to security and cloud data  Distinguishing between legitimate and malicious uses of remote management tools      Some questions we ask:        How does KQL tie into the Microsoft ecosystem, like Defender and Copilot?  What advice would you give to someone new to KQL who wants to start learning?  What is the technique we're seeing with copy-pasting malicious PowerShell?     Resources:   View Mark Morowczynski on LinkedIn  View Matt Zorich on LinkedIn  View Rod Trent on LinkedIn  View Lekshmi Vijayan on LinkedIn   View Sherrod DeGrippo on LinkedIn     Related Microsoft Podcasts:                    Afternoon Cyber Tea with Ann Johnson  The BlueHat Podcast  Uncovering Hidden Risks        Discover and follow other Microsoft podcasts at microsoft.com/podcasts     Get the latest threat intelligence insights and guidance at Microsoft Security Insider      The Microsoft Threat Intelligence Podcast is produced by Microsoft and distributed as part of N2K media network.

In this episode of the Microsoft Threat Intelligence Podcast host Sherrod DeGrippo discusses North Korean threat actors with one of our Microsoft Threat Intelligence researchers and Greg Schloemer focusing on two prominent groups: Onyx Sleet and Storm 0530. Onyx Sleet is a long-standing espionage group known for targeting defense and energy sectors, particularly in the U.S. and India. However, they’ve diversified into ransomware, using tactics like malware downloaders, zero-day vulnerabilities, and a remote access Trojan called D-Track. The conversation also touches on the use of fake certificates and the group's involvement in the software supply chain space.    In this episode you’ll learn:       The relationship between Onyx Sleet and Storm 0530  North Korea's broader strategy of using cyber-attacks and moonlighting activities  Surprising nature of recent attack chains involving vulnerability in the Chromium engine    Some questions we ask:      Does Onyx Sleet engage in cryptocurrency activities as well as traditional espionage?  How does the use of a fake Tableau software certificate fit into Onyx Sleet's attack chain?  Where does the name "Holy Ghost" come from, and why did they choose it?    Resources:   View Greg Schloemer on LinkedIn   View Sherrod DeGrippo on LinkedIn     Related Microsoft Podcasts:                    Afternoon Cyber Tea with Ann Johnson  The BlueHat Podcast  Uncovering Hidden Risks      Discover and follow other Microsoft podcasts at microsoft.com/podcasts     Get the latest threat intelligence insights and guidance at Microsoft Security Insider      The Microsoft Threat Intelligence Podcast is produced by Microsoft and distributed as part of N2K media network.

In this episode of the Microsoft Threat Intelligence Podcast host Sherrod DeGrippo is joined by Microsoft security researchers Anna Seitz and Daria Pop to discuss the latest trends in ransomware and the evolving role of AI in cyber threats. Daria Pop provides insights into the shifting tactics of Black Basta ransomware, including their use of phishing, social engineering, and remote management tools. The discussion also covers the persistence of malvertising and its challenges for defenders. Anna Seitz explores how state-sponsored threat actors, including Forest Blizzard, Emerald Sleet, and Crimson Sandstorm, are leveraging large language models (LLMs) for various malicious activities.     In this episode you’ll learn:        Why the takedown of Qakbot impacted Black Basta’s strategies  What malvertising is and why its persistence is due to the complex nature of ad traffic  How the MITRE Atlas framework assists defenders in identifying new threats    Some questions we ask:        What role does social engineering play in the campaigns involving Quick Assist?  How are North Korean threat actors like Emerald Sleep using LLMs for their campaigns?  Can you explain the changes in Black Basta’s initial access methods over the years?    Resources:   View Anna Seitz on LinkedIn   View Daria Pop on LinkedIn   View Sherrod DeGrippo on LinkedIn     Related Microsoft Podcasts:                    Afternoon Cyber Tea with Ann Johnson  The BlueHat Podcast  Uncovering Hidden Risks        Discover and follow other Microsoft podcasts at microsoft.com/podcasts     Get the latest threat intelligence insights and guidance at Microsoft Security Insider      The Microsoft Threat Intelligence Podcast is produced by Microsoft and distributed as part of N2K media network.

On this week's episode of The Microsoft Threat Intelligence Podcast, we discuss the collaborative effort between Microsoft and Fortra to combat the illegal use of cracked Cobalt Strike software, which is commonly employed in ransomware attacks.  To break down the situation, our host, Sherrod DeGrippo, is joined by Richard Boscovich, Assistant General Counsel at Microsoft, Jason Lyons, Principal Investigator with the DCU, and Bob Erdman, Associate VP Research and Development at Fortra. The discussion covers the creative use of DMCA notifications tailored by geographic region to combat cybercrime globally. The group express their optimism about applying these successful techniques to other areas, such as phishing kits, and highlight ongoing efforts to make Cobalt Strike harder to abuse.      In this episode you’ll learn:          The impact on detection engineers due to the crackdown on cracked Cobalt Strike  Extensive automation used to detect and dismantle large-scale threats  How the team used the DMCA creatively to combat cybercrime    Some questions we ask:          Do you encounter any pushback when issuing DMCA notifications?   How do you plan to proceed following the success of this operation?   Can you explain the legal mechanisms behind this take-down?    Resources:   View Jason Lyons on LinkedIn  View Bob Erdman on LinkedIn    View Richard Boscovich on LinkedIn   View Sherrod DeGrippo on LinkedIn     Related Microsoft Podcasts:                    Afternoon Cyber Tea with Ann Johnson  The BlueHat Podcast  Uncovering Hidden Risks        Discover and follow other Microsoft podcasts at microsoft.com/podcasts     Get the latest threat intelligence insights and guidance at Microsoft Security Insider      The Microsoft Threat Intelligence Podcast is produced by Microsoft and distributed as part of N2K media network.