{"course":{"slug":"ai-for-teachers","title":"AI for Teachers","url":"https://addaly.com/learn/ai-for-teachers","level":"beginner","track":"work"},"lesson":{"slug":"image-generators-and-deepfakes","title":"Image generators and deepfakes","url":"https://addaly.com/learn/ai-for-teachers/image-generators-and-deepfakes","position":45,"of":57,"minutes":10,"answer":"Fabricated intimate images of classmates are a present problem in schools, and the lesson that helps is not \"how to spot a fake\", which is becoming impossible, but what the tools do.","keyIdea":"Fabricated intimate images of classmates are a present problem in schools, and the lesson that helps is not \"how to spot a fake\", which is becoming impossible, but what the tools do, why it is a serious harm and increasingly a crime, and exactly what a student should do in the first hour if it happens to them."},"body":{"format":"markdown","text":"## Why this lesson is not optional\n\nSince 2023 there have been documented cases in Spain, the United States, South Korea, Australia and elsewhere of students using free apps to create sexual images of classmates from ordinary photographs: a school photo, a social media picture. The victims have been as young as 11. The images spread on group chats within hours. In several cases the school found out from a parent, days later, and had no plan.\n\nThis is not a future risk and it is not a rare one. A teacher of students aged 12 and up should assume it can happen in their school this year, and the lesson that helps is the one this chapter describes.\n\n## What to teach about how it works\n\nOne paragraph is enough, and it connects to the sorting game and the prediction activity. An image generator has been shown millions of pictures with descriptions, and has learned what pictures of things look like well enough to make new ones from a description. A face-swap or \"undressing\" app is the same idea pointed at one person's photograph: it does not reveal anything, it fabricates a plausible image using the pattern of thousands of other images. Nothing in the result is real. That is worth saying plainly, because victims often feel exposed as though something true had been shown, and it was not.\n\n## Why \"spot the fake\" is the wrong lesson\n\nTeachers reach for it because it is an activity. Students enjoy it. And the check question gives the result: scores near chance, and falling with every model release. Hands and text were tells two years ago; they are not now. Teaching detection by eye trains a false confidence, and the student who is confident they can spot a fake is the one who shares a real image believing it is fabricated, or a fabricated one believing it is real.\n\nThe honest replacement has three parts.\n\n**Provenance, not appearance.** Where did this come from? Who posted it first? Is there a source that can be checked? Some cameras and platforms now attach a cryptographic record of origin (the Content Credentials standard) and it is worth showing students what that looks like, while being honest that most images do not carry it and its absence proves nothing. The habit to build is asking where, not staring at fingers.\n\n**Harm, and the law.** A fabricated intimate image of a real person is a serious harm to that person, regardless of whether anyone believes it is real. Several countries have made creating or sharing such images a criminal offence since 2023, and more are following; where the subject is under 18 it falls under child sexual abuse material law in most places, with consequences that include the person who made it as a teenager. The law differs by country and is changing, and a teacher should find out what applies locally and say it clearly. Students consistently underestimate this. \"It's just a joke\" is the phrase that precedes most of the documented cases.\n\n**What to do in the first hour.** This is the part that matters most and it should be on a card:\n\n1. Do not forward it, even to show someone. Every copy is a further harm and, in many places, a further offence.\n2. Screenshot only what is needed to report (the account, the time) and then report it to a named adult at school, that day.\n3. If you are the person in the image: it is not real, it is not your fault, and the school's job is to help you. Tell a named adult. Reporting routes exist (in many countries a hotline or a platform takedown service that will remove the image from major sites) and the school should know them.\n4. The school's job in the first hour: stop the spread by direct request to students, not by announcement; contact the platforms; contact the parents of the victim first; involve the police where the law requires it; and treat the victim as a victim throughout.\n\n## The classroom conversation\n\nThis is a lesson to give with the pastoral lead, and with the room prepared: some students in it will have been affected, and some will have made or shared such an image without understanding what it was. The tone is factual and serious, not shocked. The aim is that every student leaves knowing three things (it is fabricated, it is a crime and a serious harm, and here is exactly what to do) and that the ones who have been thinking of it as a joke have had that thought quietly removed.\n\n## Boys, and the way it is framed\n\nThe documented cases are overwhelmingly boys making images of girls. A lesson that treats the whole class as potential victims misses the audience that needs it most. Say who does this and why it is not a joke to the person it is done to, and say it to the boys without accusing them. The framing that works, according to teachers who have run these sessions, is about what kind of person you are choosing to be, not about rules.\n\n## The honest limit\n\nA lesson does not stop a determined student, and the apps are free and easy. What it does is make the casual case (the one that begins as a joke in a group chat) far less likely, by removing the ignorance it depends on, and it makes the school's response fast when a case comes. The technical fixes (watermarking, provenance standards, platform detection) are improving and are not sufficient, and a teacher should not tell students that they are."},"check":{"question":"A school runs a \"spot the deepfake\" lesson in which students guess which of ten images are real. Students score about 55%. What is the honest conclusion and the better lesson?","options":[{"text":"Students need more practice at looking, and a second session with feedback on each image would raise their accuracy substantially.","correct":false,"tempting":"Detection by eye is close to chance for current tools and getting worse; more practice trains a skill that does not exist."},{"text":"Near-chance detection is the finding; teach provenance, harm and what to do instead of spotting.","correct":true},{"text":"The images chosen were too hard for a first session, and easier examples would have taught the skill before moving to difficult ones.","correct":false,"tempting":"Easier examples teach students to spot old or bad fakes, which builds false confidence about the current ones."},{"text":"Deepfakes are not yet convincing enough to matter in schools, so the lesson can wait until the technology is a real problem.","correct":false,"tempting":"Cases involving students have been documented in several countries since 2023; the problem is present, not future."}]},"prev":{"slug":"the-sorting-game-for-younger-children","title":"The sorting game for younger children","url":"https://addaly.com/learn/ai-for-teachers/the-sorting-game-for-younger-children"},"next":{"slug":"the-ai-companion-problem","title":"The AI companion problem","url":"https://addaly.com/learn/ai-for-teachers/the-ai-companion-problem"},"authorship":"Founder and series editor: Dr Mustafa Mun. Written by AI models to a written standard. Exam keys checked blind by a second model: 3,665 of 3,666 agreed. Academic review invited.","review":{"status":"Draft, open for academic review"},"cite":"Addaly (2026) 'Image generators and deepfakes', in AI for Teachers. Edited by M. Mun. Addaly. Available at: https://addaly.com/learn/ai-for-teachers/image-generators-and-deepfakes (Accessed: 13 October 2026).","licence":{"name":"CC BY-NC-SA 4.0","url":"https://creativecommons.org/licenses/by-nc-sa/4.0/"},"reader":{"schema":1,"title":"Image generators and deepfakes","courseTitle":"AI for Teachers","module":{"index":6,"slug":"teaching-ai-itself","title":"Teaching AI itself"},"answer":null,"keyIdea":"Fabricated intimate images of classmates are a present problem in schools, and the lesson that helps is not “how to spot a fake”, which is becoming impossible, but what the tools do, why it is a serious harm and increasingly a crime, and exactly what a student should do in the first hour if it happens to them.","hero":{"kind":"plate","image":null,"alt":null,"caption":null},"blocks":[{"t":"h2","id":"why-this-lesson-is-not-optional","runs":[{"text":"Why this lesson is not optional"}]},{"t":"p","runs":[{"text":"Since 2023 there have been documented cases in Spain, the United States, South Korea, Australia and elsewhere of students using free apps to create sexual images of classmates from ordinary photographs: a school photo, a social media picture. The victims have been as young as 11. The images spread on group chats within hours. In several cases the school found out from a parent, days later, and had no plan."}]},{"t":"p","runs":[{"text":"This is not a future risk and it is not a rare one. A teacher of students aged 12 and up should assume it can happen in their school this year, and the lesson that helps is the one this chapter describes."}]},{"t":"h2","id":"what-to-teach-about-how-it-works","runs":[{"text":"What to teach about how it works"}]},{"t":"p","runs":[{"text":"One paragraph is enough, and it connects to the sorting game and the prediction activity. An image generator has been shown millions of pictures with descriptions, and has learned what pictures of things look like well enough to make new ones from a description. A face-swap or “undressing” app is the same idea pointed at one person’s photograph: it does not reveal anything, it fabricates a plausible image using the pattern of thousands of other images. Nothing in the result is real. That is worth saying plainly, because victims often feel exposed as though something true had been shown, and it was not."}]},{"t":"h2","id":"why-spot-the-fake-is-the-wrong-lesson","runs":[{"text":"Why “spot the fake” is the wrong lesson"}]},{"t":"p","runs":[{"text":"Teachers reach for it because it is an activity. Students enjoy it. And the check question gives the result: scores near chance, and falling with every model release. Hands and text were tells two years ago; they are not now. Teaching detection by eye trains a false confidence, and the student who is confident they can spot a fake is the one who shares a real image believing it is fabricated, or a fabricated one believing it is real."}]},{"t":"p","runs":[{"text":"The honest replacement has three parts."}]},{"t":"p","runs":[{"text":"Provenance, not appearance.","b":true},{"text":" Where did this come from? Who posted it first? Is there a source that can be checked? Some cameras and platforms now attach a cryptographic record of origin (the Content Credentials standard) and it is worth showing students what that looks like, while being honest that most images do not carry it and its absence proves nothing. The habit to build is asking where, not staring at fingers."}]},{"t":"p","runs":[{"text":"Harm, and the law.","b":true},{"text":" A fabricated intimate image of a real person is a serious harm to that person, regardless of whether anyone believes it is real. Several countries have made creating or sharing such images a criminal offence since 2023, and more are following; where the subject is under 18 it falls under child sexual abuse material law in most places, with consequences that include the person who made it as a teenager. The law differs by country and is changing, and a teacher should find out what applies locally and say it clearly. Students consistently underestimate this. “It’s just a joke” is the phrase that precedes most of the documented cases."}]},{"t":"p","runs":[{"text":"What to do in the first hour.","b":true},{"text":" This is the part that matters most and it should be on a card:"}]},{"t":"ol","start":1,"items":[[{"text":"Do not forward it, even to show someone. Every copy is a further harm and, in many places, a further offence."}],[{"text":"Screenshot only what is needed to report (the account, the time) and then report it to a named adult at school, that day."}],[{"text":"If you are the person in the image: it is not real, it is not your fault, and the school’s job is to help you. Tell a named adult. Reporting routes exist (in many countries a hotline or a platform takedown service that will remove the image from major sites) and the school should know them."}],[{"text":"The school’s job in the first hour: stop the spread by direct request to students, not by announcement; contact the platforms; contact the parents of the victim first; involve the police where the law requires it; and treat the victim as a victim throughout."}]]},{"t":"h2","id":"the-classroom-conversation","runs":[{"text":"The classroom conversation"}]},{"t":"p","runs":[{"text":"This is a lesson to give with the pastoral lead, and with the room prepared: some students in it will have been affected, and some will have made or shared such an image without understanding what it was. The tone is factual and serious, not shocked. The aim is that every student leaves knowing three things (it is fabricated, it is a crime and a serious harm, and here is exactly what to do) and that the ones who have been thinking of it as a joke have had that thought quietly removed."}]},{"t":"h2","id":"boys-and-the-way-it-is-framed","runs":[{"text":"Boys, and the way it is framed"}]},{"t":"p","runs":[{"text":"The documented cases are overwhelmingly boys making images of girls. A lesson that treats the whole class as potential victims misses the audience that needs it most. Say who does this and why it is not a joke to the person it is done to, and say it to the boys without accusing them. The framing that works, according to teachers who have run these sessions, is about what kind of person you are choosing to be, not about rules."}]},{"t":"h2","id":"the-honest-limit","runs":[{"text":"The honest limit"}],"role":"limit"},{"t":"p","runs":[{"text":"A lesson does not stop a determined student, and the apps are free and easy. What it does is make the casual case (the one that begins as a joke in a group chat) far less likely, by removing the ignorance it depends on, and it makes the school’s response fast when a case comes. The technical fixes (watermarking, provenance standards, platform detection) are improving and are not sufficient, and a teacher should not tell students that they are."}]}],"check":{"question":[{"text":"A school runs a “spot the deepfake” lesson in which students guess which of ten images are real. Students score about 55%. What is the honest conclusion and the better lesson?"}],"options":[{"text":[{"text":"Students need more practice at looking, and a second session with feedback on each image would raise their accuracy substantially."}],"correct":false,"tempting":[{"text":"Detection by eye is close to chance for current tools and getting worse; more practice trains a skill that does not exist."}]},{"text":[{"text":"Near-chance detection is the finding; teach provenance, harm and what to do instead of spotting."}],"correct":true,"tempting":null},{"text":[{"text":"The images chosen were too hard for a first session, and easier examples would have taught the skill before moving to difficult ones."}],"correct":false,"tempting":[{"text":"Easier examples teach students to spot old or bad fakes, which builds false confidence about the current ones."}]},{"text":[{"text":"Deepfakes are not yet convincing enough to matter in schools, so the lesson can wait until the technology is a real problem."}],"correct":false,"tempting":[{"text":"Cases involving students have been documented in several countries since 2023; the problem is present, not future."}]}],"order":[3,2,1,0],"why":null},"prev":{"slug":"the-sorting-game-for-younger-children","title":"The sorting game for younger children","minutes":8},"next":{"slug":"the-ai-companion-problem","title":"The AI companion problem","minutes":9}}}