APEX

AI Agents - sounds like science fiction? You’re probably already using one! From GitHub Copilot, to Microsoft 365 Copilot, Amazon Web Services (AWS) Bedrock, and Google NotebookLM - most organization already use AI agents. The good news? They take actions. The bad news? They take actions. That’s why Gartner just published its new AI Agents new risks and threats guide. We are proud to be mentioned as one of only two recommended solutions for AI Agents security! - https://lnkd.in/dzzKxBuk AI Agents introduce NEW security risks to your organization, including being manipulated to perform malicious actions. Our recent Invisible Ghost vulnerability demonstrates this in an alarming way - https://lnkd.in/dfDFycdz Get our AI Agents risks and mitigation white paper to learn more on what you should care about when it comes to AI Agents security and how we can help you ensure AI Agents are for the benefit of your organization - https://lnkd.in/dQZaNERA Avivah Litan, Jeremy D'Hoinne, Tom Coshow

❇️Kickoff your #Blackhat this year at Topgolf Las Vegas!❇️ Join us, Cyera, Oasis Security, and Dazz, for the Scotch, Security & Scotty Cameron Kick Off Reception, meet fellow security leaders, and win a Scotty Cameron Putter. Spots are limited, register here - https://lnkd.in/dxzRwJ95

⚠️Countdown Begins: The EU AI ACT Enters into Force on August 1st, 2024⚠️ If you’re using AI chat platforms such as OpenAI ChatGPT, #Claude, or Google #Gemini, GitHub or Microsoft365 #Copilot, integrating models via API within your applications or services, or fine-tuning models/self building AI models, you will be subject to this legislation. 💸Non-compliance could lead to substantial fines, with the average fine estimated at €20 million! Use the expert calculator to assess the fine your company might face for AI Act violations - https://lnkd.in/dVGQjm_y How can you stay compliant? Apex knows regulations can be tricky, so we’ve created a practical tutorial to help you get ready for enforcement. ❇️Read the full blog by Keren Katz - https://lnkd.in/d5B_tKKS #EUAIAct #AICompliance #RegulationReady #AI #TechLaw #DataProtection

🚨GitHub Copilot vulnerability - “Invisible Ghost”🚨 Imagine your Copilot secretly spread malicious instructions throughout your codebase, jumping between files, while damaging the organization’s repository, and the integrity of your projects - that’s exactly what the Apex Research Team found 🎯 By leveraging hidden instructions, we made GitHub Copilot receive, accept, and act according to malicious prompts, bypassing its internal defense mechanisms, yielding false/malicious suggestions. By exploiting “Invisible Ghost” we were able to manipulate Copilot behavior, leading to: ⚠False suggestions and broken code outputs ⚠Malicious or vulnerable suggestions ⚠Deployment of an invisible backdoor (see our Bazinga example!) ⚠Due to its stealthiness, the exploit can propagate between users and across repos! Those can damage your code quality and security, disrupt your organization development velocity and safety, ultimately leading to a complete GitHub Copilot DOS for the organization. Don’t let your org’s security be compromised by invisible threats, check out the full report - by Fufu Shpigelman, Omer Katz, and Oren Saban - for details and mitigation strategies: https://lnkd.in/dyP9Ukk8 #AISecurityRisks #GitHubCopilot P.S. GitHub's stance? “It’s the user’s responsibility.” What’s your view on this? 🤔

Miao! 🐱 💻 Can you imagine a simple kitten would outsmart Azure AI safeguards? Our Apex AI security research team recently uncovered significant, yet easy-to-exploit, vulnerabilities in Azure’s content filtering system. The implications? Far-reaching and potentially serious for businesses relying on these safeguards. While our findings might seem playful, they highlight critical gaps in AI security that could lead to data exposure, application security risks, compliance issues, and reputational damage. Stay ahead of emerging AI security threats with Apex. Because in the world of AI, words can lead to significant consequences. #AISecurity #ContentFiltering #ResponsibleAI https://lnkd.in/drAffERc

We are excited to be featured on Gartner's Innovation Guide for Generative AI in Trust, Risk and Security Management (TRiSM)! #GenAI introduces tremendous opportunities to any organization, but also some new risks. We are proud to be the only agentless solution to remediate risks across all three #GenAI risk categories by Gartner - Content Anomalies, Data Protection, and AI Application Security. Read more about what we are building to empower you to securely drive your organization towards AI adoption -  https://www.apexhq.ai/ Thank you Avivah Litan, Jeremy D'Hoinne, and Gabriele Rigon 👏

What do you think of AI adoption in your company? Well, hackers think it’s great, but in a Matrix kind of way ⚠️ GitHub Copilot, Microsoft 365 Copilot, Notion AI, Glean, Slack AI—the AI rush is embedded into your day-to-day applications. If you have opened the door for those “assistants,” you are already at risk. "Never send a human to do a machine's job." – Agent Smith. Read more on our latest blog about AI agents by Oren Saban -their benefits and their risks ❇️ https://lnkd.in/d2KH4gDj

🎉We Did It!🎉 In just a month since launching, Apex reaches the 1,000 followers mark. Thanks to each and every one of you for your engagement and support 🤝 Looking forward to our next milestones together as we collaborate and help shape the future of AI security 🛡️

For most people in the world, writing the next post will be really hard - ChatGPT is down. For Apex customers, it's actually pretty easy, as the Apex Portal is up and running 😉 This is why vendor unlock is indispensable. Even if one vendor has an outage, you can seamlessly use the others within the same experience. Not to mention that different LLMs are suitable for different tasks... Hard to believe? Try it yourself https://lnkd.in/d2P9ZZNu

Are you familiar with the concept of Latent Space? No, it's not a physics discussion, but rather a critical cybersecurity concept that could be putting your organization at risk🌌 As AI models and applications become increasingly ubiquitous, attackers are exploiting the latent space through tactics like prompt injection and jailbreak, presenting significant security threats. In our latest article, by Apex CTO Omer Katz, we dive into the latent space, explain what actually matters to your organization, and explore how you can mitigate the risks associated with these emerging threats ✴ https://lnkd.in/dDFNKNF4