59a8 Using AI Responsibly, Safely, and Fairly | Being Clear When AI Help Should Be Disclosed, and Using AI in Ways That Help Rather Than Harm

Responsible AI use is about more than getting a technically good result. It includes being truthful about important AI assistance, avoiding deception, respecting other people's work and information, watching for unfair results, and remembering that a tool does not take responsibility for the choices made with its output.
This section prepares readers to judge not only whether AI can help with a task, but also whether the way it is being used is fair, transparent, respectful, and unlikely to cause avoidable harm.

59a8.1 Being Clear When AI Help Should Be Disclosed

Disclosure means being clear about meaningful AI assistance when another person reasonably needs to know how the work was produced. The need depends on context. A private spelling suggestion may not matter, while AI-generated research, a formal application, assessed schoolwork, professional advice, or content presented as personal experience may require clearer explanation under the rules or expectations of that setting.
Follow the relevant instructions from the school, workplace, publisher, customer, or platform rather than inventing a universal rule. When disclosure is needed, describe the assistance accurately: for example, whether AI helped brainstorm, draft, translate, edit, or generate an image. Do not claim full authorship of work you did not meaningfully create, and do not exaggerate minor assistance either. Transparency works best when it gives the audience the information needed to judge the work fairly. Clear disclosure prevents others from mistaking generated work for direct experience, research, or expertise.

59a8.2 Avoiding Deception and False Representation

AI can make text, images, audio, or other material look convincing, which creates opportunities for deception. It is misleading to use generated material to impersonate a real person, invent evidence, create a false record, or make others believe that something happened when it did not. Even a joke can cause harm if the audience cannot tell that the material is fictional or altered.
Before sharing AI-generated content, consider what a reasonable viewer is likely to believe and whether anyone could be unfairly affected. Label or explain synthetic or heavily altered material when context requires it, and avoid using AI to fabricate qualifications, endorsements, personal experiences, or proof. The responsible question is not only whether the tool can produce the content, but whether presenting it in that form would give other people a false impression that matters. If viewers could mistake generated content for a real event or statement, add clear context before sharing it.

59a8.3 Respecting Other People’s Work and Information

AI may be able to summarize, imitate, transform, or generate material based on existing work, but access to a tool does not remove the need to respect creators and information owners. Other people's writing, images, designs, data, confidential documents, and private messages may have legal, contractual, or ethical limits on how they can be used. Those limits can vary by country, platform, license, and context.
Use material you are entitled to use and follow the rules that apply to the source. When a task requires quotation, attribution, permission, or a licensed resource, handle those requirements directly rather than assuming AI makes them unnecessary. Avoid asking the tool to hide copying or pass someone else's work off as your own. Responsible assistance builds on information in a way you can explain and defend, instead of treating every accessible piece of content as free raw material. When permission is uncertain, choose material with clear rights or create original content instead.

59a8.4 Recognizing Bias and Unfair Results

AI systems can produce biased or unfair results because the patterns they learned, the data available, the design choices behind the service, or the wording of a request may reflect unequal treatment or missing perspectives. Bias can appear in recommendations, descriptions, rankings, image generation, or assumptions about what people are like. It may be obvious, but it can also appear as a subtle pattern across many results.
Look for stereotypes, unexplained differences in how groups are described, missing viewpoints, and criteria that may disadvantage people without a good reason. Try alternative wording or examples to see whether the result changes, and use independent evidence where fairness matters. For decisions affecting jobs, access, education, credit, services, or safety, human review should be especially careful. A neutral tone does not prove that an output is fair; fairness requires attention to who may be helped, ignored, or harmed by the result.

59a8.5 Keeping Human Responsibility for Important Decisions

An AI system can recommend an action, but it cannot accept responsibility for what happens after the recommendation is followed. Important decisions often involve values, duties, local rules, incomplete evidence, and consequences for several people. These are reasons to keep a responsible human involved rather than treating the system's output as the final authority.
Human responsibility includes checking critical facts, understanding the basis of the recommendation, considering alternatives, and being willing to reject the output. In workplaces or organizations, it may also mean following approval procedures and making sure a qualified person reviews decisions that require specialist judgment. Saying 'the AI decided' does not remove accountability. The tool may assist analysis, but the person or organization using it must remain able to explain why the final action was chosen. Real oversight requires a reviewer who understands the issue and has authority to change the outcome.

59a8.6 Using AI in Ways That Help Rather Than Harm

A useful AI result is not automatically a good use of AI. Consider the likely effects on other people, privacy, safety, fairness, trust, and the quality of the task being done. A shortcut that saves time for one person may create extra work, confusion, false information, or exposure for someone else. Responsible use looks at the wider outcome, not only the immediate convenience.
Before acting, ask who could benefit, who could be affected negatively, and whether the same goal can be reached with less risk. Avoid using AI to harass, deceive, discriminate, expose private information, or create unsafe instructions. When the consequences are uncertain, reduce the scale of the action, seek review, or choose a safer method. Technology is most helpful when it supports a legitimate purpose while keeping preventable harm as low as reasonably possible. A responsible user is prepared to stop using a tool for a particular task when the risks cannot be reduced to an acceptable level.