<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" version="2.0">
  <channel>
    <title>Resources</title>
    <link>https://24081737.hs-sites.com/resources</link>
    <description>Mindgard Learning Hub and Resources</description>
    <language>en</language>
    <pubDate>Fri, 06 Dec 2024 16:27:50 GMT</pubDate>
    <dc:date>2024-12-06T16:27:50Z</dc:date>
    <dc:language>en</dc:language>
    <item>
      <title>Jailbreak and Encoding Risks in Pixtral-Large-Instruct-2411</title>
      <link>https://24081737.hs-sites.com/resources/jailbreak-and-encoding-risks-in-pixtral-large</link>
      <description>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/jailbreak-and-encoding-risks-in-pixtral-large" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/ansi-escape-sequences.jpeg" alt="Jailbreak and Encoding Risks in Pixtral-Large-Instruct-2411" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;We recently tested Pixtral Large (specifically &lt;a href="https://huggingface.co/mistralai/Pixtral-Large-Instruct-2411"&gt;&lt;span&gt;Pixtral-Large-Instruct-2411&lt;/span&gt;&lt;/a&gt; announced on Nov 18, 2024) and found it exhibits vulnerabilities to multiple known jailbreak techniques, and its support for different encoding can facilitate further exploits.&lt;/p&gt;</description>
      <content:encoded>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/jailbreak-and-encoding-risks-in-pixtral-large" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/ansi-escape-sequences.jpeg" alt="Jailbreak and Encoding Risks in Pixtral-Large-Instruct-2411" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;We recently tested Pixtral Large (specifically &lt;a href="https://huggingface.co/mistralai/Pixtral-Large-Instruct-2411"&gt;&lt;span&gt;Pixtral-Large-Instruct-2411&lt;/span&gt;&lt;/a&gt; announced on Nov 18, 2024) and found it exhibits vulnerabilities to multiple known jailbreak techniques, and its support for different encoding can facilitate further exploits.&lt;/p&gt;  
&lt;img src="https://track.hubspot.com/__ptq.gif?a=24081737&amp;amp;k=14&amp;amp;r=https%3A%2F%2F24081737.hs-sites.com%2Fresources%2Fjailbreak-and-encoding-risks-in-pixtral-large&amp;amp;bu=https%253A%252F%252F24081737.hs-sites.com%252Fresources&amp;amp;bvt=rss" alt="" width="1" height="1" style="min-height:1px!important;width:1px!important;border-width:0!important;margin-top:0!important;margin-bottom:0!important;margin-right:0!important;margin-left:0!important;padding-top:0!important;padding-bottom:0!important;padding-right:0!important;padding-left:0!important; "&gt;</content:encoded>
      <category>Blog</category>
      <pubDate>Fri, 06 Dec 2024 16:25:16 GMT</pubDate>
      <author>lewis.birch@mindgard.ai (Lewis Birch)</author>
      <guid>https://24081737.hs-sites.com/resources/jailbreak-and-encoding-risks-in-pixtral-large</guid>
      <dc:date>2024-12-06T16:25:16Z</dc:date>
    </item>
    <item>
      <title>How the OWASP Top 10 Risks for LLMs Evolved from 2023 to 2025: Lessons and Implications</title>
      <link>https://24081737.hs-sites.com/resources/owasp-top-10-risks-for-llms-2025</link>
      <description>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/owasp-top-10-risks-for-llms-2025" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/OWASP%20Top%2010%20Nov%202024.png" alt="owasp top 10 for llm" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;span&gt;The release of the updated OWASP Top 10 Risks for Large Language Models (LLMs) 2025 indicates how LLMs are increasingly deployed but also highlights the emerging vulnerabilities observed by users in actual deployments.&lt;br&gt;&lt;/span&gt;&lt;/p&gt;</description>
      <content:encoded>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/owasp-top-10-risks-for-llms-2025" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/OWASP%20Top%2010%20Nov%202024.png" alt="owasp top 10 for llm" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;span&gt;The release of the updated OWASP Top 10 Risks for Large Language Models (LLMs) 2025 indicates how LLMs are increasingly deployed but also highlights the emerging vulnerabilities observed by users in actual deployments.&lt;br&gt;&lt;/span&gt;&lt;/p&gt;  
&lt;img src="https://track.hubspot.com/__ptq.gif?a=24081737&amp;amp;k=14&amp;amp;r=https%3A%2F%2F24081737.hs-sites.com%2Fresources%2Fowasp-top-10-risks-for-llms-2025&amp;amp;bu=https%253A%252F%252F24081737.hs-sites.com%252Fresources&amp;amp;bvt=rss" alt="" width="1" height="1" style="min-height:1px!important;width:1px!important;border-width:0!important;margin-top:0!important;margin-bottom:0!important;margin-right:0!important;margin-left:0!important;padding-top:0!important;padding-bottom:0!important;padding-right:0!important;padding-left:0!important; "&gt;</content:encoded>
      <category>Blog</category>
      <pubDate>Tue, 26 Nov 2024 14:22:09 GMT</pubDate>
      <author>peter@mindgard.ai (Dr. Peter Garraghan)</author>
      <guid>https://24081737.hs-sites.com/resources/owasp-top-10-risks-for-llms-2025</guid>
      <dc:date>2024-11-26T14:22:09Z</dc:date>
    </item>
    <item>
      <title>Bypassing Azure AI Content Safety Guardrails</title>
      <link>https://24081737.hs-sites.com/resources/bypassing-azure-ai-content-safety-guardrails</link>
      <description>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/bypassing-azure-ai-content-safety-guardrails" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/Cover.png" alt="Bypassing Azure AI Content Safety Guardrails" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;span&gt;In February 2024, Mindgard discovered and responsibly disclosed two security vulnerabilities within Microsoft’s &lt;a href="https://azure.microsoft.com/en-us/products/ai-services/ai-content-safety/"&gt;Azure AI Content Safety&lt;/a&gt; Service, whereby attackers could evade detection and bypass established GenAI guardrails.&lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt;</description>
      <content:encoded>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/bypassing-azure-ai-content-safety-guardrails" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/Cover.png" alt="Bypassing Azure AI Content Safety Guardrails" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;span&gt;In February 2024, Mindgard discovered and responsibly disclosed two security vulnerabilities within Microsoft’s &lt;a href="https://azure.microsoft.com/en-us/products/ai-services/ai-content-safety/"&gt;Azure AI Content Safety&lt;/a&gt; Service, whereby attackers could evade detection and bypass established GenAI guardrails.&lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt;  
&lt;img src="https://track.hubspot.com/__ptq.gif?a=24081737&amp;amp;k=14&amp;amp;r=https%3A%2F%2F24081737.hs-sites.com%2Fresources%2Fbypassing-azure-ai-content-safety-guardrails&amp;amp;bu=https%253A%252F%252F24081737.hs-sites.com%252Fresources&amp;amp;bvt=rss" alt="" width="1" height="1" style="min-height:1px!important;width:1px!important;border-width:0!important;margin-top:0!important;margin-bottom:0!important;margin-right:0!important;margin-left:0!important;padding-top:0!important;padding-bottom:0!important;padding-right:0!important;padding-left:0!important; "&gt;</content:encoded>
      <category>Blog</category>
      <pubDate>Tue, 29 Oct 2024 12:37:56 GMT</pubDate>
      <author>william.hackett@mindgard.ai (William Hackett)</author>
      <guid>https://24081737.hs-sites.com/resources/bypassing-azure-ai-content-safety-guardrails</guid>
      <dc:date>2024-10-29T12:37:56Z</dc:date>
    </item>
    <item>
      <title>The Security Strategist Podcast: Neural Networks at Risk</title>
      <link>https://24081737.hs-sites.com/resources/security-strategist-podcast-neural-networks-at-risk</link>
      <description>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/security-strategist-podcast-neural-networks-at-risk" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/Podcast%20Cover.png" alt="The Security Strategist Podcast: Neural Networks at Risk" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;em&gt;Our co-founder and CEO, &lt;a href="https://www.linkedin.com/in/pgarraghan/"&gt;Peter Garraghan&lt;/a&gt;, recently joined Paulina Ris Maya on the The Security Strategist Podcast. They discussed discussed the complexity of securing neural networks, hidden vulnerabilities within AI systems, and best AI security practices organizations should adopt.&lt;/em&gt;&lt;/p&gt;</description>
      <content:encoded>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/security-strategist-podcast-neural-networks-at-risk" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/Podcast%20Cover.png" alt="The Security Strategist Podcast: Neural Networks at Risk" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;em&gt;Our co-founder and CEO, &lt;a href="https://www.linkedin.com/in/pgarraghan/"&gt;Peter Garraghan&lt;/a&gt;, recently joined Paulina Ris Maya on the The Security Strategist Podcast. They discussed discussed the complexity of securing neural networks, hidden vulnerabilities within AI systems, and best AI security practices organizations should adopt.&lt;/em&gt;&lt;/p&gt;  
&lt;img src="https://track.hubspot.com/__ptq.gif?a=24081737&amp;amp;k=14&amp;amp;r=https%3A%2F%2F24081737.hs-sites.com%2Fresources%2Fsecurity-strategist-podcast-neural-networks-at-risk&amp;amp;bu=https%253A%252F%252F24081737.hs-sites.com%252Fresources&amp;amp;bvt=rss" alt="" width="1" height="1" style="min-height:1px!important;width:1px!important;border-width:0!important;margin-top:0!important;margin-bottom:0!important;margin-right:0!important;margin-left:0!important;padding-top:0!important;padding-bottom:0!important;padding-right:0!important;padding-left:0!important; "&gt;</content:encoded>
      <category>Podcast</category>
      <category>Blog</category>
      <pubDate>Wed, 25 Sep 2024 08:00:00 GMT</pubDate>
      <author>peter@mindgard.ai (Dr. Peter Garraghan)</author>
      <guid>https://24081737.hs-sites.com/resources/security-strategist-podcast-neural-networks-at-risk</guid>
      <dc:date>2024-09-25T08:00:00Z</dc:date>
    </item>
    <item>
      <title>AI Under Attack: Six Key Adversarial Attacks and Their Consequences</title>
      <link>https://24081737.hs-sites.com/resources/ai-under-attack-six-key-adversarial-attacks-and-their-consequences</link>
      <description>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/ai-under-attack-six-key-adversarial-attacks-and-their-consequences" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/Frame%2010083.png" alt="AI Under Attack: Six Key Adversarial Attacks and Their Consequences" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;span&gt;In this article, we explore the field of Adversarial Machine Learning, highlighting six categories of attacks that exemplify the ongoing struggle between AI security and adversarial attacks. These attack categories were chosen for discussion based on our own experiences; alongside notable research completed by others. While this overview is not exhaustive, it provides insight into the complex challenge of securing AI systems.&lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt; 
&lt;p&gt;&lt;span&gt;Adversarial Machine Learning (AML) represents a critical frontier within the cybersecurity landscape, whereby attackers craft inputs specifically designed to deceive, steal or exploit AI models, which can pose a significant threat to the integrity and reliability of AI systems. For companies, these attacks are not just a technical nuisance, but can represent severe risk to operational security, data privacy, and trustworthiness. Understanding these attacks, and how to defend against them, is paramount for developers and companies alike to ensure their AI implementations are secure and resilient.&lt;/span&gt;&lt;/p&gt;</description>
      <content:encoded>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/ai-under-attack-six-key-adversarial-attacks-and-their-consequences" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/Frame%2010083.png" alt="AI Under Attack: Six Key Adversarial Attacks and Their Consequences" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;span&gt;In this article, we explore the field of Adversarial Machine Learning, highlighting six categories of attacks that exemplify the ongoing struggle between AI security and adversarial attacks. These attack categories were chosen for discussion based on our own experiences; alongside notable research completed by others. While this overview is not exhaustive, it provides insight into the complex challenge of securing AI systems.&lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt; 
&lt;p&gt;&lt;span&gt;Adversarial Machine Learning (AML) represents a critical frontier within the cybersecurity landscape, whereby attackers craft inputs specifically designed to deceive, steal or exploit AI models, which can pose a significant threat to the integrity and reliability of AI systems. For companies, these attacks are not just a technical nuisance, but can represent severe risk to operational security, data privacy, and trustworthiness. Understanding these attacks, and how to defend against them, is paramount for developers and companies alike to ensure their AI implementations are secure and resilient.&lt;/span&gt;&lt;/p&gt;  
&lt;img src="https://track.hubspot.com/__ptq.gif?a=24081737&amp;amp;k=14&amp;amp;r=https%3A%2F%2F24081737.hs-sites.com%2Fresources%2Fai-under-attack-six-key-adversarial-attacks-and-their-consequences&amp;amp;bu=https%253A%252F%252F24081737.hs-sites.com%252Fresources&amp;amp;bvt=rss" alt="" width="1" height="1" style="min-height:1px!important;width:1px!important;border-width:0!important;margin-top:0!important;margin-bottom:0!important;margin-right:0!important;margin-left:0!important;padding-top:0!important;padding-bottom:0!important;padding-right:0!important;padding-left:0!important; "&gt;</content:encoded>
      <category>Blog</category>
      <pubDate>Thu, 29 Aug 2024 12:38:28 GMT</pubDate>
      <author>lewis.birch@mindgard.ai (Lewis Birch)</author>
      <guid>https://24081737.hs-sites.com/resources/ai-under-attack-six-key-adversarial-attacks-and-their-consequences</guid>
      <dc:date>2024-08-29T12:38:28Z</dc:date>
    </item>
    <item>
      <title>AI Red Teaming for Image Models</title>
      <link>https://24081737.hs-sites.com/resources/ai-red-teaming-for-image-models</link>
      <description>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/ai-red-teaming-for-image-models" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/Frame%2010081.png" alt="AI Red Teaming for Image Models" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;With the advent of AI and its rapid integration into almost every facet of our lives, mechanisms to ensure its security and trustworthiness are now a necessity. As outlined within NIST’s AI Risk Management Framework and practiced by leading AI-mature enterprises, red teaming AI systems is an effective approach towards achieving AI security. Such red teaming enables the identification and mitigation of security threats to AI applications, enabling organizations to accelerate their AI adoption.&lt;/p&gt; 
&lt;p&gt;We are excited to announce that starting today, Mindgard supports automated red teaming of image AI models hosted anywhere! This new capability enables AI and security teams to conduct rigorous testing of image models to identify weaknesses, biases, and potential misuse scenarios, ultimately aiming to enhance their security, reliability, and ethical performance across a wide range of applications.&lt;/p&gt;</description>
      <content:encoded>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/ai-red-teaming-for-image-models" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/Frame%2010081.png" alt="AI Red Teaming for Image Models" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;With the advent of AI and its rapid integration into almost every facet of our lives, mechanisms to ensure its security and trustworthiness are now a necessity. As outlined within NIST’s AI Risk Management Framework and practiced by leading AI-mature enterprises, red teaming AI systems is an effective approach towards achieving AI security. Such red teaming enables the identification and mitigation of security threats to AI applications, enabling organizations to accelerate their AI adoption.&lt;/p&gt; 
&lt;p&gt;We are excited to announce that starting today, Mindgard supports automated red teaming of image AI models hosted anywhere! This new capability enables AI and security teams to conduct rigorous testing of image models to identify weaknesses, biases, and potential misuse scenarios, ultimately aiming to enhance their security, reliability, and ethical performance across a wide range of applications.&lt;/p&gt;  
&lt;img src="https://track.hubspot.com/__ptq.gif?a=24081737&amp;amp;k=14&amp;amp;r=https%3A%2F%2F24081737.hs-sites.com%2Fresources%2Fai-red-teaming-for-image-models&amp;amp;bu=https%253A%252F%252F24081737.hs-sites.com%252Fresources&amp;amp;bvt=rss" alt="" width="1" height="1" style="min-height:1px!important;width:1px!important;border-width:0!important;margin-top:0!important;margin-bottom:0!important;margin-right:0!important;margin-left:0!important;padding-top:0!important;padding-bottom:0!important;padding-right:0!important;padding-left:0!important; "&gt;</content:encoded>
      <category>Product Update</category>
      <pubDate>Thu, 08 Aug 2024 13:48:59 GMT</pubDate>
      <author>stefan@mindgard.ai (Stefan Trawicki)</author>
      <guid>https://24081737.hs-sites.com/resources/ai-red-teaming-for-image-models</guid>
      <dc:date>2024-08-08T13:48:59Z</dc:date>
    </item>
    <item>
      <title>Caveat Podcast: The Startup Leading AI Security in the UK with Mindgard's CEO</title>
      <link>https://24081737.hs-sites.com/resources/caveat-podcast-the-startup-leading-ai-security-in-the-uk-mindgard</link>
      <description>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/caveat-podcast-the-startup-leading-ai-security-in-the-uk-mindgard" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/peter%20frame%20(2).png" alt="Dr. Peter Garraghan on the Caveat Podcast by Cyberwire" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;em&gt;Our co-founder and CEO, &lt;a href="https://www.linkedin.com/in/pgarraghan/"&gt;Peter Garraghan&lt;/a&gt;, recently joined &lt;a href="https://www.linkedin.com/in/dave-bittner-27231a4/"&gt;Dave Bittner &lt;/a&gt;on the &lt;a href="https://thecyberwire.com/podcasts/caveat"&gt;Caveat Podcast&lt;/a&gt; on &lt;a href="https://thecyberwire.com/"&gt;The Cyber Wire&lt;/a&gt;. They discussed the UK's recently published AI security guidelines and the recommendations Peter made for addressing cybersecurity risks in AI.&lt;/em&gt;&lt;/p&gt;</description>
      <content:encoded>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/caveat-podcast-the-startup-leading-ai-security-in-the-uk-mindgard" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/peter%20frame%20(2).png" alt="Dr. Peter Garraghan on the Caveat Podcast by Cyberwire" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;em&gt;Our co-founder and CEO, &lt;a href="https://www.linkedin.com/in/pgarraghan/"&gt;Peter Garraghan&lt;/a&gt;, recently joined &lt;a href="https://www.linkedin.com/in/dave-bittner-27231a4/"&gt;Dave Bittner &lt;/a&gt;on the &lt;a href="https://thecyberwire.com/podcasts/caveat"&gt;Caveat Podcast&lt;/a&gt; on &lt;a href="https://thecyberwire.com/"&gt;The Cyber Wire&lt;/a&gt;. They discussed the UK's recently published AI security guidelines and the recommendations Peter made for addressing cybersecurity risks in AI.&lt;/em&gt;&lt;/p&gt;  
&lt;img src="https://track.hubspot.com/__ptq.gif?a=24081737&amp;amp;k=14&amp;amp;r=https%3A%2F%2F24081737.hs-sites.com%2Fresources%2Fcaveat-podcast-the-startup-leading-ai-security-in-the-uk-mindgard&amp;amp;bu=https%253A%252F%252F24081737.hs-sites.com%252Fresources&amp;amp;bvt=rss" alt="" width="1" height="1" style="min-height:1px!important;width:1px!important;border-width:0!important;margin-top:0!important;margin-bottom:0!important;margin-right:0!important;margin-left:0!important;padding-top:0!important;padding-bottom:0!important;padding-right:0!important;padding-left:0!important; "&gt;</content:encoded>
      <category>Podcast</category>
      <category>Blog</category>
      <pubDate>Fri, 26 Jul 2024 18:22:20 GMT</pubDate>
      <author>haralds.gz@mindgard.ai (Haralds Gabrans Zukovs)</author>
      <guid>https://24081737.hs-sites.com/resources/caveat-podcast-the-startup-leading-ai-security-in-the-uk-mindgard</guid>
      <dc:date>2024-07-26T18:22:20Z</dc:date>
    </item>
    <item>
      <title>Bypassing AI-Driven Deepfake Detection via Evasion Attacks</title>
      <link>https://24081737.hs-sites.com/resources/bypassing-ai-driven-deepfake-detection-via-evasion-attacks</link>
      <description>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/bypassing-ai-driven-deepfake-detection-via-evasion-attacks" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/william%20article.png" alt="Bypassing AI-Driven Deepfake Detection via Evasion Attacks by William Hackett" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;span&gt;Deepfakes enable the production of highly realistic videos where individuals appear to say or do things they never actually did. This technology has found applications in various fields, from entertainment and marketing to more controversial uses such as &lt;/span&gt;&lt;a href="https://newlinesmag.com/spotlight/ai-and-deepfakes-played-a-big-role-in-indias-elections/"&gt;&lt;span&gt;political campaigns&lt;/span&gt;&lt;/a&gt;&lt;span&gt;, &lt;/span&gt;&lt;a href="https://www.bbc.co.uk/news/technology-60780142"&gt;&lt;span&gt;disinformation&lt;/span&gt;&lt;/a&gt;&lt;span&gt;, and &lt;/span&gt;&lt;a href="https://www.cnbc.com/2024/05/28/deepfake-scams-have-looted-millions-experts-warn-it-could-get-worse.html"&gt;&lt;span&gt;fraud&lt;/span&gt;&lt;/a&gt;&lt;span&gt;. &lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt; 
&lt;p&gt;&lt;span&gt;To combat their misuse, deepfake detection techniques have been developed that use AI models to classify whether a video or image is genuine or a deepfake. These AI models are designed to identify manipulated content by learning features and patterns prevalent within deepfake videos and images that are not always perceptible to the human eye. &lt;/span&gt;&lt;/p&gt; 
&lt;p&gt;&lt;span&gt;Such techniques detect subtle inconsistencies or artifacts left over from the deepfake generation process. These detection models play a crucial role in maintaining the integrity of information, as well as protecting individuals from digital forgery and targeted cyber harm campaigns.&lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt; 
&lt;p&gt;&lt;span&gt;However, as deepfake detection technology evolves, so too do the techniques used to evade it. An emerging threat within the research and industry landscape is the development of sophisticated attacks aimed specifically at bypassing AI-driven deepfake detection. These evasion tactics can make it extremely difficult for AI models to identify when content has been manipulated.&lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt; 
&lt;p&gt;&lt;span&gt;In this blog, we will dive into how these evasion attacks can outsmart current deepfake detection systems. We will make use of &lt;/span&gt;&lt;a href="https://sandbox.mindgard.ai/"&gt;&lt;span&gt;Mindgard AI Security Labs&lt;/span&gt;&lt;/a&gt;&lt;span&gt; to automate and rapidly red team deepfake detection models against a range of evasion attacks.&lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt;</description>
      <content:encoded>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/bypassing-ai-driven-deepfake-detection-via-evasion-attacks" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/william%20article.png" alt="Bypassing AI-Driven Deepfake Detection via Evasion Attacks by William Hackett" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;span&gt;Deepfakes enable the production of highly realistic videos where individuals appear to say or do things they never actually did. This technology has found applications in various fields, from entertainment and marketing to more controversial uses such as &lt;/span&gt;&lt;a href="https://newlinesmag.com/spotlight/ai-and-deepfakes-played-a-big-role-in-indias-elections/"&gt;&lt;span&gt;political campaigns&lt;/span&gt;&lt;/a&gt;&lt;span&gt;, &lt;/span&gt;&lt;a href="https://www.bbc.co.uk/news/technology-60780142"&gt;&lt;span&gt;disinformation&lt;/span&gt;&lt;/a&gt;&lt;span&gt;, and &lt;/span&gt;&lt;a href="https://www.cnbc.com/2024/05/28/deepfake-scams-have-looted-millions-experts-warn-it-could-get-worse.html"&gt;&lt;span&gt;fraud&lt;/span&gt;&lt;/a&gt;&lt;span&gt;. &lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt; 
&lt;p&gt;&lt;span&gt;To combat their misuse, deepfake detection techniques have been developed that use AI models to classify whether a video or image is genuine or a deepfake. These AI models are designed to identify manipulated content by learning features and patterns prevalent within deepfake videos and images that are not always perceptible to the human eye. &lt;/span&gt;&lt;/p&gt; 
&lt;p&gt;&lt;span&gt;Such techniques detect subtle inconsistencies or artifacts left over from the deepfake generation process. These detection models play a crucial role in maintaining the integrity of information, as well as protecting individuals from digital forgery and targeted cyber harm campaigns.&lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt; 
&lt;p&gt;&lt;span&gt;However, as deepfake detection technology evolves, so too do the techniques used to evade it. An emerging threat within the research and industry landscape is the development of sophisticated attacks aimed specifically at bypassing AI-driven deepfake detection. These evasion tactics can make it extremely difficult for AI models to identify when content has been manipulated.&lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt; 
&lt;p&gt;&lt;span&gt;In this blog, we will dive into how these evasion attacks can outsmart current deepfake detection systems. We will make use of &lt;/span&gt;&lt;a href="https://sandbox.mindgard.ai/"&gt;&lt;span&gt;Mindgard AI Security Labs&lt;/span&gt;&lt;/a&gt;&lt;span&gt; to automate and rapidly red team deepfake detection models against a range of evasion attacks.&lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt;  
&lt;img src="https://track.hubspot.com/__ptq.gif?a=24081737&amp;amp;k=14&amp;amp;r=https%3A%2F%2F24081737.hs-sites.com%2Fresources%2Fbypassing-ai-driven-deepfake-detection-via-evasion-attacks&amp;amp;bu=https%253A%252F%252F24081737.hs-sites.com%252Fresources&amp;amp;bvt=rss" alt="" width="1" height="1" style="min-height:1px!important;width:1px!important;border-width:0!important;margin-top:0!important;margin-bottom:0!important;margin-right:0!important;margin-left:0!important;padding-top:0!important;padding-bottom:0!important;padding-right:0!important;padding-left:0!important; "&gt;</content:encoded>
      <category>Blog</category>
      <pubDate>Fri, 19 Jul 2024 07:50:49 GMT</pubDate>
      <author>william.hackett@mindgard.ai (William Hackett)</author>
      <guid>https://24081737.hs-sites.com/resources/bypassing-ai-driven-deepfake-detection-via-evasion-attacks</guid>
      <dc:date>2024-07-19T07:50:49Z</dc:date>
    </item>
    <item>
      <title>Audio-Based Jailbreak Attacks on Multi-Modal LLMs</title>
      <link>https://24081737.hs-sites.com/resources/audio-based-jailbreak-attacks-on-multi-modal-llms</link>
      <description>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/audio-based-jailbreak-attacks-on-multi-modal-llms" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/Frame%2010073.png" alt="Audio-Based Jailbreak Attacks on Multi-Modal LLMs" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;Large language models (LLMs) are increasingly integrating various data types, such as text, images, and audio, into a single framework, model or system known as multi-modal LLMs. These are capable of understanding and generating human-like responses, making them invaluable in numerous applications ranging from customer service to content creation.&lt;/p&gt;</description>
      <content:encoded>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/audio-based-jailbreak-attacks-on-multi-modal-llms" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/Frame%2010073.png" alt="Audio-Based Jailbreak Attacks on Multi-Modal LLMs" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;Large language models (LLMs) are increasingly integrating various data types, such as text, images, and audio, into a single framework, model or system known as multi-modal LLMs. These are capable of understanding and generating human-like responses, making them invaluable in numerous applications ranging from customer service to content creation.&lt;/p&gt;  
&lt;img src="https://track.hubspot.com/__ptq.gif?a=24081737&amp;amp;k=14&amp;amp;r=https%3A%2F%2F24081737.hs-sites.com%2Fresources%2Faudio-based-jailbreak-attacks-on-multi-modal-llms&amp;amp;bu=https%253A%252F%252F24081737.hs-sites.com%252Fresources&amp;amp;bvt=rss" alt="" width="1" height="1" style="min-height:1px!important;width:1px!important;border-width:0!important;margin-top:0!important;margin-bottom:0!important;margin-right:0!important;margin-left:0!important;padding-top:0!important;padding-bottom:0!important;padding-right:0!important;padding-left:0!important; "&gt;</content:encoded>
      <category>Blog</category>
      <pubDate>Mon, 08 Jul 2024 13:22:34 GMT</pubDate>
      <author>lewis.birch@mindgard.ai (Lewis Birch)</author>
      <guid>https://24081737.hs-sites.com/resources/audio-based-jailbreak-attacks-on-multi-modal-llms</guid>
      <dc:date>2024-07-08T13:22:34Z</dc:date>
    </item>
    <item>
      <title>Introducing Mindgard MITRE ATLAS™ Adviser</title>
      <link>https://24081737.hs-sites.com/resources/introducing-mindgard-mitre-atlas-adviser</link>
      <description>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/introducing-mindgard-mitre-atlas-adviser" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/001_%20Blog%20Header%20Design.png" alt="Introducing Mindgard MITRE ATLAS Adviser Article" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;span&gt;As AI gets infused in every software application that powers our personal and corporate lives, ensuring their safety and security is more crucial than ever. One effective way to achieve this is through AI red teaming - a process that simulates adversarial attacks on AI systems to identify potential vulnerabilities.&lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt;</description>
      <content:encoded>&lt;div class="hs-featured-image-wrapper"&gt; 
 &lt;a href="https://24081737.hs-sites.com/resources/introducing-mindgard-mitre-atlas-adviser" title="" class="hs-featured-image-link"&gt; &lt;img src="https://24081737.hs-sites.com/hubfs/001_%20Blog%20Header%20Design.png" alt="Introducing Mindgard MITRE ATLAS Adviser Article" class="hs-featured-image" style="width:auto !important; max-width:50%; float:left; margin:0 15px 15px 0;"&gt; &lt;/a&gt; 
&lt;/div&gt; 
&lt;p&gt;&lt;span&gt;As AI gets infused in every software application that powers our personal and corporate lives, ensuring their safety and security is more crucial than ever. One effective way to achieve this is through AI red teaming - a process that simulates adversarial attacks on AI systems to identify potential vulnerabilities.&lt;/span&gt;&lt;span&gt;&amp;nbsp;&lt;/span&gt;&lt;/p&gt;  
&lt;img src="https://track.hubspot.com/__ptq.gif?a=24081737&amp;amp;k=14&amp;amp;r=https%3A%2F%2F24081737.hs-sites.com%2Fresources%2Fintroducing-mindgard-mitre-atlas-adviser&amp;amp;bu=https%253A%252F%252F24081737.hs-sites.com%252Fresources&amp;amp;bvt=rss" alt="" width="1" height="1" style="min-height:1px!important;width:1px!important;border-width:0!important;margin-top:0!important;margin-bottom:0!important;margin-right:0!important;margin-left:0!important;padding-top:0!important;padding-bottom:0!important;padding-right:0!important;padding-left:0!important; "&gt;</content:encoded>
      <category>Product Update</category>
      <pubDate>Wed, 26 Jun 2024 15:19:43 GMT</pubDate>
      <author>nipun@mindgard.ai (Nipun Gupta)</author>
      <guid>https://24081737.hs-sites.com/resources/introducing-mindgard-mitre-atlas-adviser</guid>
      <dc:date>2024-06-26T15:19:43Z</dc:date>
    </item>
  </channel>
</rss>
