Exam vocabulary
AI+ Security PathTerms™
33 plain-language terms mapped for AI+ Security Level I. Lists are Accumentum study aids based on our glossary and PathKey™ coverage, not a vendor blueprint reprint.
Showing 33 terms in a learn-these-first order.
- Data Preparation for LLMsData preparation for LLMs is cleaning and organizing data to train AI models effectively, ensuring accuracy and reducing bias.
- LLM DeploymentLLM deployment is making a trained large language model available for use in production environments, integrating it into applications and workflows.
- Model poisoningModel poisoning corrupts AI training data, causing flawed predictions and impacting system integrity.
- System promptA system prompt provides instructions to an AI model, guiding its behavior and output for more relevant and controlled interactions.
- Prompt injectionPrompt injection is a security flaw in AI where attackers manipulate input prompts to make AI systems perform unintended actions or reveal sensitive data.
- Adversarial AI Risk MitigationDefending AI systems against malicious attacks that aim to deceive, manipulate, or compromise their functionality.
- Adversarial exampleAn adversarial example is an input designed to trick an AI model into making a wrong prediction, posing a security risk.
- AI AccountabilityAI Accountability ensures AI systems are responsible, ethical, and transparent, with clear responsibility for outcomes and impacts.
- AI agentAn AI agent is a self-directed software program that perceives, decides, and acts to achieve goals, forming the basis of intelligent automation.
- AI Attack ConceptsAI Attack Concepts are adversarial techniques targeting AI systems, crucial for secure development and operation.
- AI+ Security Level IAI+ Security Level I is a professional certification topic covered in Accumentum training catalogs and PathKey™ guides.
- AI Risk ScenariosAI Risk Scenarios are hypothetical situations outlining potential negative outcomes from AI systems, guiding proactive mitigation.
- AI SecurityAI Security protects AI systems from threats and ensures their ethical, responsible use, covering models, data, and infrastructure.
- AI Security Program DevelopmentAI Security Program Development is the strategic integration of AI into an organization's security posture to enhance threat detection and response.
- AI Security Threat LandscapeThe AI Security Threat Landscape maps out risks and vulnerabilities unique to artificial intelligence systems and their applications.
- AI TRiSMAI TRiSM ensures AI systems are trustworthy, secure, and ethically managed throughout their lifecycle.
- EvalEval assesses AI model performance and quality, ensuring reliability and safety through rigorous testing and metrics.
- Foundation modelA foundation model is a large AI model trained on diverse data, capable of many tasks with little fine tuning.
- Foundation ModelsFoundation models are large, versatile AI models trained on massive datasets, adaptable to many tasks with minimal fine-tuning.
- GroundingGrounding anchors AI responses to factual context, preventing hallucinations and ensuring reliable outputs in professional AI applications.
- GuardrailsGuardrails are rules that control AI behavior, ensuring safe, ethical, and aligned outputs.
- HallucinationAI hallucinations are when generative models confidently produce false or nonsensical information, impacting real-world applications and trust.
- Introduction to AI GovernanceIntroduction to AI Governance is an artificial intelligence concept used across modern IT and certification study.
- JailbreakingJailbreaking bypasses AI safety filters to elicit restricted responses, a key concept in AI security and certification studies.
- LLM AttacksMalicious techniques exploiting vulnerabilities in Large Language Models to cause unintended or harmful behavior.
- Model extractionModel extraction is an adversarial attack to steal AI model parameters or functionality by querying it and observing outputs.
- Red teaming AIRed teaming AI proactively finds vulnerabilities in AI systems by simulating attacks, ensuring safer and more reliable AI.
- Responsible AI Governance FrameworksResponsible AI Governance Frameworks ensure ethical, safe, and value-aligned AI development and deployment through structured policies and oversight.
- Responsible AI LeadershipResponsible AI Leadership ensures ethical and accountable development, deployment, and management of AI systems.
- Responsible AI Security PracticesEnsuring AI systems are secure, safe, and trustworthy throughout their lifecycle.
- secure AI systemsSafeguarding AI systems against threats and vulnerabilities throughout their lifecycle, from design to deployment.
- Securing AI AlgorithmsSecuring AI algorithms protects AI models and data from attacks and vulnerabilities, ensuring reliable and ethical operation.
- Vulnerabilities in AI UsageWeaknesses in AI systems that can be exploited, leading to negative outcomes. Crucial for secure AI development.
Keep going
Open ReadyScore™ Cards on the PathTerms™ hub, then check ReadyScore™ when you want a scored readiness read.