Pangea

Recent research reveals how AI agents misuse web tools, solve CAPTCHAs, exfiltrate data and even poison LLMs with just a few samples. From ForcedLeak in Salesforce AgentForce to CometJacking in Perplexity’s browser and Ollama server exposure, our next webinar on Oct 28 @ 12 ET breaks down how these exploits work and what you can do to stop them. See live demos, new research and mitigation strategies. Save your spot: https://lnkd.in/eFT-fFuZ #AISecurity #CyberSecurity #AIDR #RedTeam Joey Melo Pangea

From Netskope’s IPO, to the acquisitions of Wiz, CyberArk, Nozomi Networks, and most recently, Ballistic portfolio company Pangea, what’s the state of cybersecurity exits? Ballistic GP Barmak Meftah and TAG Infosphere CEO Edward Amoroso break down what these moves mean for the industry and what may be ahead for IPOs and M&A. 🎧 Tune in: https://lnkd.in/gEZKjZ2A

In David Strom's latest piece for CSO Online, "5 steps for deploying agentic AI red teaming," he covers the evolving complexity of AI attack surfaces. Pangea AI Red Teaming Specialist Joey Melo shares his perspective on the unpredictable nature of these AI vulnerabilities, particularly agentic prompt injection: "The same payload will fail 99 times and work once, but in an unexpected way." Read more to learn how AI red teaming is democratizing the future of security testing: https://lnkd.in/ejNn8PzA #AgenticAI #AISecurity #CyberSecurity #AIDR #PromptInjection

In addition to focusing on enabling a security operations center (SOC) fundamentally powered by agents—the “agentic SOC”—CrowdStrike is also “pioneering” a new category in AI security: “AI detection and response,” in part through the just-announced deal to acquire AI guardrails startup Pangea, CrowdStrike co-founder and CEO George Kurtz said during his keynote at #FalCon2025 this week in Las Vegas. CrowdStrike is also taking the initial steps toward developing security-focused artificial general intelligence, or security AGI, according to Kurtz. Kurtz made the remarks as CrowdStrike also announced an array of new cybersecurity offerings related to AI and agents, including the Falcon Agentic Security Platform, a new set of AI agents through the launch of its Agentic Security Workforce offering and Charlotte AI AgentWorks, a no-code platform for building, testing, deploying and orchestrating security agents. Kyle Alspach breaks down five key AI takeaways from Kurtz at Fal.Con 2025:

CrowdStrike to Acquire Pangea to Secure Every Layer of Enterprise AI https://lnkd.in/dWCEQtbK #Acquisition #AIDetectionandResponse #AIsecurity #CIOFirst #CrowdStrike #Cybersecurity #EnterpriseAI #News #Pangea

CrowdStrike is acquiring Pangea. Here’s why it matters. The agent has replaced the endpoint. For years, security was built on the idea that every device had a human behind it. That idea is now obsolete. The next wave of “users” inside your company will not be people at all. They will be autonomous AI agents. They will outnumber humans by orders of magnitude. And they will operate with both autonomy and privileged access. That creates a new attack surface. Not devices. Not entitlements. But "conversations." Let's take the one attack vector every AI vendor uses - indirect prompt injection: a malicious instruction is hidden in the white text of an email. The recipient never sees it. But an AI assistant does. It reads the prompt, follows the command, and quietly performs actions on objectives using the email recipient's permissions. The system was not hacked. It convinced itself to break. This mirrors a classic intelligence dilemma: how do you protect a system when your own messengers can be turned against you? The difference today is scale. Shadow AI agents are ones you did not deploy, do not track, and cannot govern (not easily, not yet), and they already exists inside most enterprises. CrowdStrike’s move is not a reaction to today’s prompt injection headlines. It is a strategic bet on where the real fight will take place. The real vulnerability is not that AI agents can be hacked, but that they can be *persuaded*. The future of security is not about keeping bad actors out. It is about making sure the good actors you rely on cannot be turned against you.

Attending #FalCon2025? Swing on by the Pangea booth to learn how AI Detection & Response (AIDR) equips security teams with the detection and response needed for AI. Pangea's AI Security Platform helps security teams protect employee AI use and ship secure AI apps fast with the industry's most comprehensive set of guardrails. Secure AI from cloud to code! Explore and get a demo: https://lnkd.in/e4xGzRAW #CyberSecurity #AISecurity #LLMSecurity #AIDR

Today’s the day! LLM vulnerabilities are surfacing almost weekly. Join our first session in the series to see the latest exploits and learn the strategies to defend. Don’t miss it, register now below. #AIDR #AISecurity #CyberSecurity #ShadowAI #LegalPwn