About Pangea
Pangea sells an AI Security Platform that Pangea says helps security teams secure AI from cloud to code. Pangea's AI Security Platform helps security teams protect employee AI use and ship secure AI apps fast.[5]
Pangea says it offers unified visibility into all AI activity, from workforce tool usage to in-app AI. Pangea says security teams get the oversight to proactively manage risk, so that every prompt, response, and action is traceable and auditable. Pangea says organizations can protect workforce AI use, enable secure AI innovation, and ship secure AI applications faster with AI Detection & Response and its set of AI guardrails.
Pangea stops LLM security threats ranging from prompt injection to sensitive data leakage. The platform deploys via gateways, browser plugins, and into applications with just a few lines of code. Pangea provides a common AI security framework that can intercept AI usage anywhere in the enterprise from cloud workloads to employee devices to applications. Collectors can be deployed at any location in the enterprise with just a few clicks or a few lines of code.[5]
Pangea AI Detection & Response, called AIDR, is deployed via collectors that integrate with existing tools and infrastructure. Pangea says AIDR gives security teams a holistic picture of AI activity across their organization, AI threats and policy violations, and an audit trail to investigate them. Pangea collectors integrate with infrastructure from cloud workloads, to employee endpoints, to homegrown AI applications. Pangea AIDR console provides a central command center where teams can see real time AI activity and configure policy and response actions. Teams can validate and investigate AI alerts, which can be exported to SIEM environments.[5]
Organizations can track GenAI application and model coverage, identify shadow AI, enforce approved model policies, measure token consumption and support compliance reporting. Pangea says AIDR detects and blocks AI-specific attacks like prompt injection, jailbreak attempts, and malicious content insertion in real-time in over 100 languages. Pangea also lets organizations create custom access rules based on users, LLM or application attributes, and add prompt detectors to redact PII, block malicious prompts, and report on filtered topics.[5]
Pangea's data redaction and file sanitization prevented sensitive data disclosure to backend AI models, according to a published case study. Pangea Labs has published a research report on an attack vector abusing legal disclaimers to trigger prompt injections capable of bypassing malware detection. Pangea researchers have also developed a prompt injection taxonomy. Oliver Friedrichs is Founder and CEO of Pangea, and he discusses the four primary attack vectors targeting enterprise AI applications on the Detection at Scale podcast. Pangea is located at 636 Ramona St, Palo Alto, CA 94301.[3]
Pangea on the record