What the Department of Labor's New AI Literacy Standard Means for Your Team

In February 2026, the U.S. Department of Labor published its first Artificial Intelligence Literacy Framework, naming five specific skills every worker needs and seven principles for teaching them. For any team trying to move past ad hoc AI experimentation, the framework answers a question that has lingered since generative tools arrived at work: what does being AI literate actually require?

What Did the Department of Labor Actually Publish?

On February 13, 2026, the department's Employment and Training Administration released an AI Literacy Framework through Training and Employment Notice 07-25, directing it to state workforce agencies, American Job Centers, and community colleges nationwide. The notice frames the framework as a resource for program design, and its content areas translate just as well to a private employer building internal training as to a public workforce office.

This matters for AI University's audience because it is the first time a federal agency has put a name and a structure around AI literacy for the general workforce, rather than for developers or data scientists. Teams no longer have to guess at what "AI skilled" should mean when they design training or evaluate a course, and AI University's approach to workforce AI education was built around exactly that gap between using a tool and being able to direct and judge it.

The U.S. framework also lands inside a broader international push to define AI competence rather than assume it. The OECD's ongoing work on AI and skills has spent the past two years pressing member countries to move past general statements about "AI adoption" and toward specific, teachable competencies. The Department of Labor's five content areas are the first attempt by a U.S. agency to answer that call in plain, workforce-facing terms.

What Are the Five AI Literacy Content Areas?

The framework organizes AI literacy into five content areas that build on one another, moving from concept to judgment. A learner who can work through all five has gone from operating an AI tool to genuinely directing and evaluating it.

• Understanding AI Principles: knowing at a working level how models generate output, where their limits sit, and why two prompts can produce very different results.
• Exploring AI Uses: recognizing which tasks in a specific job or workflow are worth handing to AI, and which are not.
• Directing AI Effectively: writing prompts and instructions precise enough to get a usable result on the first or second try.
• Evaluating AI Outputs: catching errors, fabricated details, and bias before a result reaches a client, a record, or a decision.
• Using AI Responsibly: applying privacy, disclosure, and data handling practices appropriate to the task and the employer.

How Does the Framework Say Training Should Be Delivered?

The framework pairs those five content areas with seven delivery principles, and the first is direct: AI literacy is built through hands-on use, not lecture-style instruction. A single recorded webinar on how to use a chatbot satisfies none of the five areas above on its own.

The remaining principles, summarized from the department's notice and analysis from Ogletree Deakins, describe what that hands-on training should look like in practice:

• Embed learning in context: tie training to a worker's actual job, industry, or existing program, such as an apprenticeship or a technical curriculum.
• Build complementary human skills: pair AI training with judgment, communication, and review skills that AI cannot replace.
• Address prerequisites to AI literacy: confirm basic digital access and comfort before layering AI skills on top.
• Create pathways for continued learning: treat AI literacy as an ongoing track, not a single session.
• Prepare enabling roles: give managers and trainers, not just individual contributors, the skills to support others.
• Design for agility: build training that can be updated as tools and use cases change.

What Does Directing and Evaluating AI Actually Look Like at Work?

The framework's language stays general on purpose, but the skills it describes show up as very specific moments in a workday. A marketing analyst asking a model to draft a client-facing report is exercising "directing," and whether that draft is usable on the first try depends on how precisely they scoped the request, the audience, and the format.

A customer service lead reading an AI-drafted response before it goes out is exercising "evaluating," and the skill is not spotting bad grammar, it is catching a policy detail the model invented because it sounded plausible. A finance team member using AI to summarize a vendor contract is exercising both at once, directing the tool toward the clauses that matter and evaluating the summary against the source document before anyone relies on it. None of that is taught by a single demo of a chatbot's interface.

Does This Create a New Legal Requirement for Employers?

No. The framework is voluntary and does not carry a compliance deadline or reporting obligation, a point legal analysis of the notice has been careful to clarify. It is guidance for program design, not a mandate with an enforcement mechanism attached.

The absence of a mandate is what makes the framework useful rather than something to comply with. It gives teams a credible, government-authored definition of AI literacy to build against, without the pressure that would push training toward checking a box instead of building a skill.

The Department Also Launched a Free Course. Here Is What It Does and Does Not Do

In March 2026, the department paired the framework with "Make America AI-Ready," a free course delivered entirely by text message, built with the learning platform Arist and covered in detail by HR Dive. Workers text "READY" to a short code and receive daily lessons they can finish in about ten minutes a day over a week, covering what AI can and cannot do, how to give it effective instructions, how to evaluate what it produces, and how to use it responsibly.

The design choice matters. A text-based course reaches someone without a laptop or reliable broadband, which is precisely the accessibility gap SHRM's coverage of the course highlighted. What it is not built to do is verify anyone's competence at a specific job, or give an employer a way to show a client or a regulator that a team meets a defined standard. A seven-day introduction and a demonstrated, role-specific skill are two different things, and an organization that wants the second needs more than the free course was designed to provide.

Where Most Workplace AI Training Still Falls Short

Set the five content areas next to what passes for AI training at most organizations today and the gap is obvious. Many teams have had a single tool demo, a company-wide email about a new AI assistant, or a short recorded session, and none of those experiences touch directing, evaluating, or responsible use with any rigor. Employees are using AI daily while the organization has no real way to say whether they are doing it well.

That gap is a measurement problem as much as a training problem. An organization can assume its people are AI literate because they use AI tools, or it can find out. Coverage of the framework in higher education has already noted that colleges are using it to redesign entire courses rather than bolt on a single AI module, and workplace teams have the same option available to them.

Turning the Framework Into a Learning Plan

Start by finding out where your team actually stands against the five content areas before building anything new. An AI readiness assessment gives you that baseline, so training time goes toward the areas where a team is genuinely weak, whether that is directing AI with precise instructions or catching a fabricated detail in an output, instead of repeating an introductory demo everyone has already sat through.

From there, design training the way the framework describes it: tied to real job tasks, delivered hands-on, and treated as a track rather than an event. A marketing team practicing on its own campaign briefs and a finance team practicing on its own reporting templates will both build the five skills faster than either would from a generic course built for neither.

Do not skip the principle most teams overlook: preparing enabling roles. A manager who has never directed or evaluated an AI output personally cannot coach a direct report through a bad one, and a training plan that only reaches individual contributors will stall the first time someone needs a second opinion. Managers and team leads should go through the same hands-on practice, on their own work, before they are asked to review anyone else's.

Finally, verify the result instead of assuming it. Attendance at a training session, or a completed text-message course, says nothing on its own about whether someone can direct AI effectively or catch a bad output before it matters. A structured AI certification path gives individuals and teams a way to demonstrate, not just claim, that they have built the specific competencies the department just named as the national standard.

Next Step

The federal government has now defined what AI literacy means for the workforce. The next decision is whether your team's training actually builds it or just gestures at it. If you want help auditing your current training against the five content areas and building a plan to close the gaps, book a strategy call.