AI literacy refers to the foundational knowledge and skills required to understand, use, and critically evaluate artificial intelligence (AI) systems. It typically includes basic awareness of how AI works, where it is used, its benefits and limitations, and its ethical and social implications. AI literacy emphasizes competent and informed use of AI tools. In most frameworks, AI literacy focuses on:

• Understanding what AI is and is not
• Knowing how AI systems are trained and deployed
• Recognizing bias, limitations, and ethical concerns
• Using AI responsibly and appropriately
• AI literacy largely assumes that users are interacting with AI as intended.

Adversarial literacy is an emerging concept that builds on AI literacy—but goes further. It emphasizes the ability to actively interrogate, challenge, and stress-test AI systems. It involves understanding how AI systems can be manipulated, misled, or produce harmful outcomes—and developing the skills to recognize, expose, and respond to those vulnerabilities. Rather than asking “How do I use AI well?”, adversarial literacy asks:

“How can AI fail, be exploited, or mislead—and how do I detect that?”

Adversarial literacy includes competencies such as:

• Recognizing adversarial inputs, prompts, or examples designed to manipulate AI behavior
• Probing AI systems to reveal bias, hallucinations, or unsafe outputs
• Evaluating AI responses for reliability, intent, and context sensitivity
• Engaging in ethical “red-teaming” or adversarial questioning to surface system weaknesses
• Understanding that AI systems are contestable, not authoritative

The term adversarial literacy is increasingly used by scholars and others to describe a paradigm shift, i.e. the emergence of several AI-driven shifts:

• AI systems are no longer passive tools: Generative and predictive systems actively shape information environments—and can be manipulated intentionally or unintentionally.
• Trust is no longer sufficient: AI outputs may sound authoritative while being incorrect, biased, or strategically misleading, requiring users to adopt a skeptical stance.
• Power and agency are at stake: Adversarial literacy equips individuals (earners, workers, citizens) to resist the over-reliance on opaque systems and reclaim human judgment.
• Education must move from use to critique: Teaching people how to prompt (chat with) AI is insufficient; they must also learn how to break, test, and question it safely and ethically.

See Glossary Term: Opacity (in Artificial Intelligence) | Opaque AI | Learn & Work Ecosystem Library

See Glossary Term: Information Literacy | Learn & Work Ecosystem Library

See Topic Brief: Converging Terms:  Digital Literacy & Information Literacy | Learn & Work Ecosystem Library