FAQs

AI Compliance Monitoring ensures that AI systems align with internal policies, legal standards, and ethical guidelines. It involves tracking usage, logging decisions, detecting violations, and generating reports—giving organizations the visibility and control needed to meet frameworks like GDPR, NIST AI RMF, and the EU AI Act.

The EU AI Act is a first-of-its-kind regulation governing the development and use of artificial intelligence in the European Union. It classifies AI systems by risk level—banning some, regulating others—and mandates transparency, human oversight, and security for high-risk AI applications.

Generative AI Security refers to protecting systems built on models like GPT, Claude, and open-source LLMs from misuse, exploitation, or data compromise. It includes defenses against prompt injection, hallucinations, jailbreaks, data leakage, and supply chain vulnerabilities—ensuring safe use of AI in applications, agents, and APIs.

The NIST AI Risk Management Framework (AI RMF) is a U.S. government-led guide that helps organizations develop and deploy trustworthy AI. It outlines practices for identifying, assessing, managing, and governing AI risks, with emphasis on security, privacy, fairness, transparency, and accountability.

The OWASP Top 10 for LLMs is a widely adopted list of the most critical security risks affecting large language model (LLM) applications. It includes threats like prompt injection, data leakage, and insecure plugins—helping developers and security teams prioritize defenses across the generative AI lifecycle.

An AI Firewall is a runtime control layer that monitors and filters inputs and outputs between users and a large language model. It prevents prompt injection, jailbreaks, data leaks, and unsafe content generation—protecting applications that embed LLMs in production environments.

LLM Observability gives teams visibility into how large language models (LLMs) behave in production. It includes monitoring prompts, outputs, performance, risk signals, and user interactions—helping organizations detect failures, prevent abuse, and improve reliability of generative AI systems.

AI Supply Chain Risk refers to vulnerabilities introduced through third-party models, data, and software used to build and deploy AI systems. These risks include malicious components, license violations, or unvetted model behavior—posing security, legal, and operational threats to AI-powered applications.

An AI Bill of Materials (AI-BOM) is a detailed inventory of all components used to build and deploy an AI system—such as models, datasets, libraries, agents, and APIs. It provides transparency, traceability, and accountability across the AI supply chain for security, compliance, and governance.

AI Drift Detection monitors changes in model behavior, data inputs, or output quality over time. It identifies when an AI model starts producing inaccurate, inconsistent, or biased results—helping organizations maintain model reliability, prevent performance degradation, and ensure alignment with intended use and regulatory expectations.

Prompt Logging records all user inputs (prompts) and model-generated outputs within an AI system. It enables auditing, debugging, and compliance by maintaining a traceable record of interactions—helping teams detect misuse, resolve incidents, and verify that AI behavior aligns with policies and expectations.

AI Runtime Monitoring continuously observes prompts, model outputs, and interactions as AI systems run in production. It detects threats like prompt injection, data leakage, and jailbreaks in real time—providing visibility and control over how AI is used across live applications and environments.

Model hallucination refers to an AI model generating false, misleading, or nonsensical information with high confidence. These fabricated outputs can misinform users or disrupt downstream processes—especially in LLM-powered applications that synthesize content, code, or responses without verifying factual accuracy.

Model Risk Management (MRM) is the process of identifying, assessing, and mitigating the risks associated with AI and machine learning models. It ensures that models are accurate, secure, and aligned with regulatory expectations—reducing the chance of failure, misuse, or unintended consequences.

AI Governance is the framework of policies, processes, and controls used to ensure responsible and compliant use of AI. It addresses data integrity, model explainability, accountability, and regulatory alignment—ensuring that AI systems are secure, trustworthy, and aligned with organizational goals and ethical standards.

Shadow AI refers to artificial intelligence tools, models, or agents adopted by employees or teams without approval from security, IT, or governance groups. Like shadow IT, it introduces unmanaged risks—including data leakage, policy violations, and exposure to unvetted models or APIs—outside of centralized visibility and control.

A shadow model is an AI model deployed or developed without visibility or approval from the security team. Like shadow IT, shadow models bypass governance, introducing unmanaged risk—from unvetted datasets to insecure endpoints or unclear licensing—and often go undetected in modern development environments.

Jailbreaking is the process of manipulating an AI model—especially an LLM—to bypass safety filters or moderation policies. These attacks can force the model to generate content it’s designed to block, such as offensive language, harmful instructions, or private information.

Prompt Injection is a type of adversarial attack where users manipulate inputs to trick an AI model into ignoring its intended instructions. These attacks can bypass safety rules, hijack model behavior, or extract sensitive information—posing serious security risks in LLM-powered applications and agents.

PointGuard AI was founded by CEO Pravin Kothari, Chief Product & Security Officer Mali Gorantla, and Chief Technology Officer Warlu Kothapalli. Together, they bring decades of leadership across cloud security, product innovation, and AI-driven engineering.