Media & Information Literacy Grade 7: Deepfakes and Synthetic Media

Grade 7 Media & Information Literacy Study Guide: Deepfakes and Synthetic Media

1. The Driving Question

"If a video of the president saying something outrageous goes viral, but no one was actually in the room when it was filmed—how do you know if it’s real? And if you can’t tell, how do you decide what to believe, share, or even laugh at?"

This isn’t just about "fake news." It’s about whether your brain can trust your own eyes—and what happens when technology learns to lie better than humans can spot the lie.

2. The Core Idea — Built, Not Listed

Imagine you’re scrolling through your phone and see a video of your favorite singer performing a song they never actually sang. The voice sounds perfect, the face moves naturally, and the background looks like a real concert. But the whole thing was generated by an AI trained on hours of their interviews and performances. This is a deepfake: synthetic media created using artificial intelligence to manipulate or fabricate audio, video, or images so convincingly that it’s hard to tell it’s not real.

Deepfakes work by feeding an AI thousands of examples of a person’s face, voice, or mannerisms. The AI learns patterns—like how someone’s mouth moves when they say "hello" or how their eyebrows raise when they’re surprised—and then predicts what they would look or sound like saying something new. The scarier part? The tools to make deepfakes are getting easier to use, and the results are getting harder to detect. This isn’t just a problem for celebrities or politicians; it could be used to bully classmates, scam grandparents, or even frame someone for a crime they didn’t commit.

The real puzzle isn’t just how deepfakes are made—it’s why they’re dangerous. When we can’t trust what we see, we have to rely on other clues: where the video came from, who shared it, and whether it matches other reliable sources. But even then, the line between "real" and "fake" gets blurrier every day.

Key Vocabulary: - Deepfake Definition: Synthetic media (video, audio, or images) created or altered using AI to make it appear real, often by swapping faces, voices, or actions. Example: A TikTok video where a teenager’s face is seamlessly edited onto a dancer’s body in a way that looks like they performed the moves themselves. Note: In high school and beyond, deepfakes raise legal questions about consent, copyright, and even free speech—like whether a deepfake of a politician counts as satire or defamation.

• Generative AI Definition: A type of artificial intelligence that creates new content (text, images, audio, video) based on patterns it learns from existing data. Example: An app that turns a selfie into a "painting" in the style of Van Gogh, or a chatbot that writes a poem in the voice of Shakespeare. Note: In college, you’ll study how generative AI can both amplify creativity (e.g., designing new medicines) and undermine trust (e.g., flooding the internet with convincing fake reviews).

• Misinformation Definition: False or misleading information shared without the intent to deceive (e.g., someone sharing a deepfake because they think it’s real). Example: A friend posts a video of a shark swimming in a flooded street after a hurricane, believing it’s real footage—when it’s actually a deepfake from a movie. Note: In media studies, misinformation is often contrasted with disinformation, which is deliberately false and spread to cause harm.

• Digital Watermark Definition: A hidden signal embedded in media (like a video or image) to prove its origin or detect tampering. Example: A news organization adds an invisible watermark to its videos so that if someone edits them into a deepfake, the watermark breaks, revealing the manipulation. Note: In cybersecurity, watermarks are also used to track leaks (e.g., if a company’s confidential document is leaked, the watermark can identify who shared it).

3. Assessment Translation

How This Appears on State Assessments (Grade 7): Deepfakes and synthetic media are tested in media literacy or digital citizenship units, often through: - Multiple-choice questions (e.g., identifying red flags in a suspicious video). - Short-answer prompts (e.g., explaining how to verify a source). - Evidence-based writing (e.g., arguing whether a deepfake should be allowed in a specific context, like a political ad).

Distractor Patterns in Multiple Choice: - Over-reliance on "gut feeling": "If it looks real, it probably is" (wrong—deepfakes are designed to look real). - Assuming all fakes are obvious: "Deepfakes always have glitches" (wrong—some are nearly flawless). - Confusing misinformation with disinformation: "All fake videos are created to trick people" (wrong—some are shared by people who believe they’re real).

What a Proficient Response Looks Like: - Multiple-choice: Selects answers that emphasize verification (e.g., "Check the source’s reputation" or "Look for inconsistencies in lighting/shadows"). - Short answer: Names specific tools (e.g., reverse image search, fact-checking websites like Snopes) and explains why they work. - Writing prompt: Uses evidence (e.g., "The video’s audio doesn’t match the lip movements, which is a common deepfake flaw") and considers multiple perspectives (e.g., "While deepfakes can be used for harm, they’re also used in movies to de-age actors").

Model Proficient Response (Short Answer): Prompt: "A classmate shares a video of the principal announcing a snow day, but you’re not sure it’s real. What steps would you take to verify it before telling your friends?"

Response:
1. Check the source: Is the video from the school’s official social media or a trusted news outlet? If it’s from a random account, it’s more likely to be fake.
2. Look for inconsistencies: Do the principal’s words match their usual way of speaking? Are there weird glitches, like unnatural blinking or blurry edges?
3. Use tools: Do a reverse image search on the video’s thumbnail to see if it’s been posted elsewhere. Check fact-checking sites like PolitiFact or Snopes.
4. Ask an adult: If I’m still unsure, I’d ask a teacher or parent to help verify it. Sharing a fake snow day video could cause chaos at school!

Why this is proficient: - Names specific verification steps (not just "be careful"). - Explains why each step matters (e.g., "random account" = less trustworthy). - Shows awareness of consequences (e.g., "cause chaos").

4. Mistake Taxonomy

Mistake 1: Assuming "Real" = "True" Prompt: "A video shows a celebrity saying something offensive. How can you tell if it’s real or a deepfake?" Common Wrong Response: "If it looks real, it probably is. Celebrities say dumb stuff all the time." Why It Loses Credit: - Ignores the possibility of manipulation. - Doesn’t use any verification strategies. Correct Approach: - Check the source: Is the video from the celebrity’s verified account or a reputable news outlet? If not, it’s suspect. - Look for context: Does the statement match what the celebrity has said before? Deepfakes often show people saying things out of character. - Use tools: Run the video through a deepfake detector like Microsoft Video Authenticator or check fact-checking sites.

Mistake 2: Overlooking Small Details Prompt: "Describe two visual clues that might indicate a video is a deepfake." Common Wrong Response: "The video is blurry" or "The person’s face looks weird." Why It Loses Credit: - Too vague—deepfakes can be high-quality. - Doesn’t name specific flaws (e.g., lighting, shadows, or audio sync). Correct Approach: - Lighting and shadows: Deepfakes often have inconsistent lighting (e.g., a person’s face is lit from the left, but their shadow falls to the right). - Blinking and breathing: Some deepfakes have unnatural blinking patterns or stiff facial movements because the AI struggles with subtle details. - Audio sync: The voice might not match the lip movements perfectly, especially with certain sounds (e.g., "b" or "p" sounds).

Mistake 3: Trusting "Official" Sources Without Question Prompt: "A news channel tweets a video of a politician admitting to a crime. Should you share it? Why or why not?" Common Wrong Response: "Yes, because it’s from a news channel, so it must be real." Why It Loses Credit: - Assumes all news outlets verify content equally. - Doesn’t consider that even reputable sources can be fooled or hacked. Correct Approach: - Check for verification: Did the news channel confirm the video’s authenticity? Look for phrases like "verified by our team" or "source confirmed." - Cross-reference: Do other trusted news outlets report the same thing? If it’s a big story, multiple sources should cover it. - Consider the motive: Could this video be used to manipulate public opinion? Deepfakes are often shared to provoke emotional reactions (e.g., anger or fear).

5. Connection Layer

1. Within Media Literacy-Algorithmic Bias Understanding deepfakes clarifies how social media algorithms work. If a deepfake video goes viral, platforms like TikTok or YouTube might push it to more people because it’s engaging (even if it’s fake). This shows how algorithms prioritize attention over truth—a problem that also affects things like targeted ads or political propaganda.

2. Across Subjects-Forensic Science (Science) Deepfake detection uses the same logic as forensic analysis. Just like forensic scientists look for inconsistencies in crime scene evidence (e.g., fingerprints that don’t match), deepfake detectors look for flaws in videos (e.g., unnatural eye movements). Both fields rely on pattern recognition and skepticism of "obvious" answers.

3. Outside School-Online Gaming and Avatars Deepfake technology is already in games like Fortnite or Roblox. Some games let players create hyper-realistic avatars using AI, which raises questions: If your avatar looks like a celebrity, is that okay? What if someone uses your face without permission? This is the same ethical dilemma as deepfakes, but in a space where kids already spend hours.

6. The Stretch Question

"If a deepfake video of you went viral—showing you doing something embarrassing or illegal—what legal or technological protections should exist to help you? Should there be a law against creating deepfakes without consent, even if they’re ‘just for fun’?"

Pointer Toward the Answer: - Legal side: Some states (like California and Texas) already have laws against deepfakes used in elections or pornography, but these laws are new and untested. Should they apply to all deepfakes, or just the harmful ones? What counts as "harm"? - Technological side: Companies like Adobe and Microsoft are developing tools to "watermark" AI-generated content, but these aren’t foolproof. Should platforms like Instagram or YouTube be required to label deepfakes? - Ethical side: Even "harmless" deepfakes (e.g., putting your face on a superhero) can feel violating. How do we balance creativity with consent? Would you want a law that lets you sue someone for using your face without permission?

This isn’t just a hypothetical—it’s a debate happening right now in courts and tech companies. The answer might shape how we use (and trust) the internet for decades.