L1
Β·
Quiz
Β·
Lab
L2
Β·
Quiz
Β·
Lab
L3
Β·
Quiz
Β·
Lab
L4
Β·
Quiz
Β·
Lab
Module Test
Lesson 1 Β· Module 2

Reverse Image Search & Visual Verification

Old photos, new contexts β€” how recycled images become misinformation's most durable weapon
How can a photograph taken years ago be made to seem like breaking news?

Hours after the 2013 Boston Marathon bombing, a photograph of a missing child circulated on Twitter and Facebook with the caption identifying him as a victim. The child was Missing-Persons case Sunil Tripathi β€” but he was not at the marathon. The image had been repurposed from an entirely different missing-person campaign. His family saw their son's face attached to an atrocity he had nothing to do with. Reverse image search, had anyone applied it immediately, would have traced the photo's origin in minutes.

Why Images Lie More Effectively Than Text

Photographs carry an automatic presumption of truth. When a compelling image accompanies a claim, readers process the claim faster, believe it more strongly, and share it more readily β€” a phenomenon researchers call the photo-as-proof heuristic. Misinformation actors exploit this by attaching real photographs to false contexts: a flood image from 2011 labeled as 2023, a protest from one country labeled as another, or a celebrity photograph with fabricated captions.

Three categories of visual misinformation dominate fact-checker caseloads: recycled images (real photos, wrong date or location), manipulated images (cropped, colorized, or composited), and synthetic images (AI-generated content). Lesson 3 covers AI-generated media specifically; here we focus on the first two, which remain far more common in the wild.

The Core Tool: Reverse Image Search

Reverse image search allows you to submit an image (or image URL) and retrieve pages where that image β€” or visually similar images β€” appears. Three engines dominate professional verification work:

Tool 01
Google Images
Broadest index. Right-click any image in Chrome β†’ "Search image with Google." Identifies near-duplicates across billions of indexed pages. Best for finding the earliest known appearance of an image.
Tool 02
TinEye
Specialized reverse image engine. Sorts results by "oldest" to find the first indexed appearance β€” critical for proving an image predates a claimed event. Free tier covers most fact-checking needs.
Tool 03
Yandex Images
Russian-developed engine with superior facial recognition and strong Cyrillic-language source indexing. Often finds matches that Google and TinEye miss, particularly for images originating in Eastern Europe or Central Asia.
Tool 04
InVID / WeVerify
Browser extension (Chrome/Firefox) built for journalists. Splits videos into keyframes for reverse searching, reads video metadata, and integrates with multiple reverse search engines simultaneously. Free from the WeVerify consortium.

The SIFT Protocol Applied to Images

The SIFT method β€” Stop, Investigate the source, Find better coverage, Trace claims β€” maps directly onto visual verification. When you encounter a compelling image:

  • 1
    Stop before sharing. Emotional resonance is a red flag, not a green light. The more an image moves you, the more carefully you should examine it.
  • 2
    Right-click or use InVID to run the image through at least two reverse search engines. Note the earliest dated appearance you can find.
  • 3
    Check metadata. Platforms strip EXIF data on upload, but images downloaded from original sources often retain GPS coordinates, camera model, and original timestamp. Tools: Jeffrey's Exif Viewer, ExifTool.
  • 4
    Cross-reference the claimed location with Google Maps Street View or satellite imagery. Shadow direction, architecture, vegetation, and road markings can confirm or refute a claimed geography.
  • 5
    Look for established fact-checker coverage. Snopes, PolitiFact, AFP Fact Check, Reuters Fact Check, and First Draft have likely already investigated widely circulated images.
Documented Case β€” 2020 Australian Wildfires

During the catastrophic 2019–20 Australian bushfires, a photograph labeled "Australia on fire" went globally viral. Fact-checkers at AAP FactCheck and AFP traced it via TinEye to a NASA satellite image of California wildfires from 2018. The misattribution did not diminish the real severity of the Australian crisis β€” but it demonstrated how emotional urgency short-circuits verification instincts even among experienced news consumers.

Geolocation: Matching Images to Places

Geolocation is the process of determining where a photograph or video was taken using visual cues in the image itself. It requires no special software β€” only systematic observation and access to mapping tools. The Bellingcat investigative collective has published detailed geolocation guides and demonstrated the method in dozens of conflict-zone investigations, including tracking Russian military hardware movements in Ukraine by matching road signs, bridge structures, and tree lines to satellite imagery.

Key visual anchors for geolocation: road signs and house numbers, distinctive architecture or landmarks, shadow length and direction (which can be used with sun-position calculators to verify the time of day and season), vegetation species (which narrow geographic region), and military or emergency vehicle markings.

Key Terms
Recycled ImageA genuine photograph reposted with a false date, location, or caption to fit a current narrative.
EXIF DataExchangeable Image File Format metadata embedded in digital photos β€” includes timestamp, GPS, and camera settings. Often stripped by social platforms.
GeolocationDetermining the precise location where an image was captured using visual cues cross-referenced with mapping tools.
TinEyeReverse image search engine that can sort results by earliest indexed date β€” essential for establishing an image's provenance timeline.
Quiz Β· Lesson 1

Reverse Image Search & Visual Verification

Three questions β€” select the best answer for each.
1. Which reverse image search tool is most useful for finding the earliest indexed appearance of an image?
Correct. TinEye's "oldest" sort is specifically designed to establish when an image first appeared online β€” the key question when determining whether a photo predates the event it's claimed to show.
Not quite. While Google Images has a larger index, TinEye's ability to sort results by oldest date is the critical feature for establishing a photo's provenance timeline.
2. During the 2013 Boston Marathon bombing, misinformation spread identifying an innocent person as a suspect. What type of visual misinformation was primarily involved?
Correct. A real photograph of Sunil Tripathi β€” taken for an unrelated missing-person campaign β€” was recycled and falsely identified as a bombing suspect. No manipulation of the image itself was required; only the false caption.
Not quite. The image was a genuine photograph placed in a false context β€” the most common form of visual misinformation, requiring no technical manipulation at all.
3. Shadow direction and length in a photograph can be used by fact-checkers to verify what?
Correct. By combining shadow direction and length with a sun-position calculator and the claimed location, investigators can verify or refute the stated time and date of a photograph β€” a technique used extensively in conflict-zone verification.
Not quite. Shadow analysis with a sun-position calculator lets investigators estimate time of day and season β€” useful for verifying or refuting a claimed timestamp.
Lab Β· Lesson 1

Image Verification Practice

Work through visual verification scenarios with your AI lab partner.

Scenario: You're a newsroom fact-checker

A social media post claims a photograph shows flooding in Jakarta, Indonesia, from "last week." You suspect the image may be recycled. Your AI lab partner will walk you through the verification workflow β€” ask it anything about how to investigate this image, what tools to use, what to look for, and how to document your findings.

Suggested opener: "Walk me through the first steps I'd take to verify whether this flood photo is really from Jakarta last week." β€” or ask any question about visual verification tools and methods.
Visual Verification Lab
AI Lab Partner
Ready to work through visual verification. You've got a flood photograph someone claims is from Jakarta last week β€” what's your first move? Ask me about any tool or step, and I'll walk you through the professional workflow.
Lesson 2 Β· Module 2

Fact-Checking Organizations & Lateral Reading

Why opening new tabs beats reading more carefully β€” the counterintuitive skill of professional verification
What do professional fact-checkers do that ordinary readers almost never do?

Stanford researchers asked three groups β€” historians, professional fact-checkers, and undergraduate students β€” to evaluate the credibility of websites. The historians and students carefully read within the websites they were given, scrutinizing design, "About" pages, and citations. The professional fact-checkers did something entirely different: they immediately opened new browser tabs to search for what others said about each source. The fact-checkers were faster and dramatically more accurate. The researchers named this behavior lateral reading.

What Lateral Reading Actually Means

Lateral reading means leaving a website almost immediately and searching for what credible third parties say about it β€” rather than reading the website itself more carefully. This sounds obvious, but it runs against the instinct most people develop in school, where careful close reading is rewarded. Online, close reading plays into the hands of sophisticated misinformation sites that invest heavily in appearing credible from within.

The practical workflow: when you encounter an unfamiliar source making a significant claim, open a new tab, type the source name into a search engine, and scan the first few results. Wikipedia, journalism watchdogs, academic critiques, and news stories about the organization appear quickly. You are not trying to find the "truth" about the claim yet β€” you are establishing whether the source is trustworthy enough to read further.

The Major Fact-Checking Organizations

Over 300 fact-checking organizations operate globally, indexed by the Duke Reporters' Lab. The most cited in English-language verification work:

US β€” General
PolitiFact
Pulitzer Prize–winning outlet using a "Truth-O-Meter" scale from True to Pants on Fire. Strong on US political claims. Operated by the Poynter Institute since 2018.
US β€” General
Snopes
Oldest English-language fact-checking site (1994). Covers viral claims, urban legends, and political misinformation. Particularly strong on image and social media claim verification.
US β€” Political
FactCheck.org
Nonpartisan project of the Annenberg Public Policy Center at University of Pennsylvania. Focuses on US political claims; launched SciCheck for science misinformation in 2015.
International
AFP Fact Check
Global operation running fact-checks in 26 languages across 120+ countries. Particularly strong on image verification for conflict zones and health misinformation. Partners with Facebook's third-party fact-checking program.
International
Reuters Fact Check
Launched 2020. Leverages Reuters' global reporter network for rapid, multi-continent verification. Especially strong on breaking news claims where speed matters.
Standards Body
IFCN (Poynter)
The International Fact-Checking Network sets ethical standards for fact-checkers. Signatories commit to nonpartisanship, transparency, and corrections policies. Maintains a public list of verified signatories.

Reading Fact-Check Ratings Correctly

Fact-check verdicts are more nuanced than simple true/false. PolitiFact's "Half True" rating means a claim contains accurate elements but omits critical context that would change its meaning. AFP's "Misleading" rating often covers technically accurate statements weaponized through selective framing. Understanding rating systems prevents the mistake of treating a "Mostly True" verdict as full endorsement.

A critical caveat: absence of a fact-check is not evidence of truth. Fact-checking organizations are small and must triage. The vast majority of circulating misinformation has never been reviewed. A claim that hasn't been checked is not thereby verified β€” it simply hasn't been checked.

Case: The IFCN Code of Principles

When Facebook launched its third-party fact-checking program in 2016 after criticism over misinformation during the US election, it required partner organizations to be certified IFCN signatories. The IFCN Code requires: a commitment to nonpartisanship, transparent sourcing, transparent funding, transparent methodology, and an open corrections policy. Checking whether a fact-checking outlet is an IFCN signatory is itself a quick lateral-reading move that filters out politically motivated "fact-checkers" operating under that label.

Building a Personal Verification Toolkit

Professional fact-checkers maintain bookmarked sets of starting points. A functional personal toolkit includes: a reverse image search extension (InVID/WeVerify), three fact-checking bookmarks (Snopes, PolitiFact or AFP, and a science-specific checker like FullFact for health claims), a media bias reference (AllSides or Ad Fontes Media Bias Chart), and an archived search tool (Wayback Machine / archive.org) for retrieving deleted content.

The Wayback Machine deserves special mention: when a source deletes a page after publishing misinformation, the archived version often remains. Fact-checkers routinely use it to document claims that have since been removed, establishing a permanent record.

Key Terms
Lateral ReadingLeaving a website to search for what others say about it, rather than reading the site itself more carefully. Identified as the primary skill distinguishing professional fact-checkers from non-experts.
IFCNInternational Fact-Checking Network β€” the Poynter Institute body that certifies fact-checking organizations against ethical standards including nonpartisanship and transparency.
Wayback MachineInternet Archive tool at archive.org that stores historical snapshots of web pages β€” essential for retrieving deleted misinformation content.
Quiz Β· Lesson 2

Fact-Checking Organizations & Lateral Reading

Three questions β€” select the best answer for each.
1. What did the Stanford History Education Group's 2016 study find about professional fact-checkers' behavior when evaluating a website?
Correct. This is lateral reading β€” leaving the site almost immediately to check external sources. Fact-checkers were faster and more accurate than historians and students who read carefully within the site.
That describes what the historians and students did β€” and they were less accurate. Fact-checkers instead opened new tabs to search for what credible third parties said about the source.
2. A fact-check verdict of "Misleading" typically means:
Correct. "Misleading" often covers technically accurate statistics or quotes that are stripped of context, cherry-picked, or framed to imply something the underlying facts don't support.
Not quite. "Misleading" typically applies to content that is technically accurate but deceptively framed β€” selective use of real data is one of the most common examples.
3. Which tool would a fact-checker use to retrieve a webpage that has since been deleted by its publisher?
Correct. The Internet Archive's Wayback Machine stores historical snapshots of web pages, allowing fact-checkers to document claims that publishers later delete.
Not quite. The Wayback Machine at archive.org stores historical snapshots of web pages β€” the standard tool for retrieving deleted content.
Lab Β· Lesson 2

Lateral Reading & Source Evaluation

Practice evaluating unfamiliar sources using the lateral reading method.

Scenario: Evaluating an unfamiliar news outlet

You encounter a health article from a website called "NaturalHealthInsider.net" claiming a government agency suppressed a study on vaccine side effects. You've never heard of this outlet. Your AI lab partner will help you apply lateral reading and other source evaluation techniques to assess its credibility.

Suggested opener: "How do I quickly evaluate whether NaturalHealthInsider.net is a credible source?" β€” or ask about lateral reading, media bias tools, IFCN certification, or how to find who funds a website.
Source Evaluation Lab
AI Lab Partner
You've got an article from an unfamiliar health site making a serious claim. Where do you start? Ask me about lateral reading, how to check funding and ownership, IFCN certification, or anything about evaluating source credibility.
Lesson 3 Β· Module 2

Detecting AI-Generated Content

Synthetic images, voice clones, and deepfakes β€” the new verification frontier and why tools alone aren't enough
When an image was never a photograph to begin with, what traces does it leave?

On March 22, 2023, a series of photorealistic images depicting former President Donald Trump being arrested by police officers spread across Twitter, Telegram, and news aggregators. The images were created by Eliot Higgins of Bellingcat using Midjourney v5, posted with a label identifying them as AI-generated β€” and then extensively recirculated without that label by other accounts. Multiple news outlets ran queries to their newsrooms asking for authentication. The images had never represented a real event, yet they provoked genuine public confusion about whether an arrest had occurred.

The Detection Gap

AI image generation has outpaced detection technology. In 2023, researchers at the University of California, Berkeley, and elsewhere demonstrated that leading AI detectors β€” including those embedded in major platforms β€” had false-positive rates that made them unreliable for operational fact-checking. A human rights organization using an AI detector could incorrectly flag a genuine atrocity photograph as synthetic, with serious real-world consequences.

The honest assessment from the fact-checking community: no single AI detection tool is reliable enough to use as a sole verification method. Detection tools are useful as a triage signal β€” raising a flag for closer investigation β€” but not as a final verdict. The verification workflow must combine technical signals with contextual analysis.

Current Detection Tools and Their Limits

Image Detection
Hive Moderation
API-based classifier returning probability scores for AI generation. Among the most accurate in independent benchmarks as of 2024. Used by newsrooms and platforms. Output is a probability, not a verdict β€” treat scores below 95% with caution.
Image Detection
AI or Not (Illuminarty)
Free web tool for image classification. Useful for quick triage. Accuracy varies significantly by image generation model and post-processing. Most reliable for images from older generation models; less reliable for latest diffusion outputs.
Text Detection
GPTZero
Trained to detect LLM-generated text using perplexity and burstiness measures. Significant false-positive rates for non-native English writers, academic text, and technical writing. Should never be used as sole evidence of AI authorship.
Provenance Standard
C2PA / Content Credentials
Coalition for Content Provenance and Authenticity β€” an open technical standard for cryptographically signing images at point of creation with metadata about origin. Adobe, Microsoft, Leica, Nikon, and others are signatories. Verify at contentcredentials.org/verify.

Visual Artifacts: What to Look For

Current AI image generators leave characteristic artifacts that trained eyes can learn to spot β€” though this list will become obsolete as models improve. As of 2024, common signals include:

  • β†’
    Hand and finger anomalies. Diffusion models historically struggled with hands β€” extra fingers, merged digits, impossible joint angles. This has improved in newer models but remains a useful check.
  • β†’
    Text within images. AI-generated text in signs, newspapers, and backgrounds is frequently garbled, misspelled, or nonsensical. Real photographs contain legible text.
  • β†’
    Background consistency. Crowd scenes often show repeated faces, anatomically impossible backgrounds, or symmetric patterns that don't occur naturally.
  • β†’
    Lighting inconsistency. AI images sometimes show incompatible light sources β€” shadows falling in different directions, or specular highlights that don't match the apparent light environment.
  • β†’
    Ear and hair detail. Ears are frequently malformed or asymmetric; hair near faces often blends unnaturally with the background.
Documented Case β€” Deepfake Audio, Slovakia 2023

Days before Slovakia's September 2023 parliamentary election, an audio clip circulated on Facebook appearing to be a recording of Progressive Slovakia leader Michal Ε imečka discussing how to rig the election by buying votes from the Roma community. AFP Fact Check and DFRLab analyzed the audio and identified characteristics consistent with AI voice synthesis β€” unnatural prosody, background noise inconsistencies, and phrasing atypical of Ε imečka's documented speaking style. The clip could not be definitively proven synthetic with available tools, but multiple signals pointed strongly in that direction. Ε imečka's party narrowly lost the election. The case became a reference point in EU discussions about AI regulation ahead of elections.

The Contextual Verification Approach

Because technical detection is unreliable, experienced fact-checkers lean heavily on contextual signals. The key questions for any suspicious image or video: Does corroborating evidence exist from independent sources? Is the claim being amplified by accounts with patterns of sharing synthetic content? Does the content appear at a suspiciously convenient political moment? Can the claimed event be verified through other media?

The First Draft organization, before its 2023 closure, documented a principle they called verification through triangulation: if an event happened, multiple independent witnesses with different devices and angles will have captured it. A single perfect image of a major event with no corroborating footage is a red flag regardless of what detection tools say.

Key Terms
C2PACoalition for Content Provenance and Authenticity β€” an open standard for cryptographically embedding provenance metadata into images and videos at the point of creation.
Detection GapThe disparity between the sophistication of AI content generation and the reliability of tools designed to detect it β€” currently, generators outpace detectors.
TriangulationThe verification principle that genuine events produce multiple independent records β€” absence of corroborating evidence from other sources is a red flag for synthetic or staged content.
Quiz Β· Lesson 3

Detecting AI-Generated Content

Three questions β€” select the best answer for each.
1. What is the primary limitation of current AI image detection tools for operational fact-checking?
Correct. High false-positive rates mean genuine photographs can be incorrectly flagged as AI-generated β€” with potentially serious consequences in contexts like human rights documentation. Detection tools should be triage signals, not final verdicts.
Not quite. The core problem is false-positive rates β€” real photographs being incorrectly classified as synthetic. This makes detection tools useful for triage but not reliable as sole evidence.
2. The C2PA (Coalition for Content Provenance and Authenticity) standard addresses misinformation by:
Correct. C2PA embeds a cryptographic chain of custody into media files at creation, allowing anyone to verify the origin and edit history of an image or video. This is a provenance approach rather than a detection approach.
Not quite. C2PA works at the point of creation β€” embedding cryptographic provenance metadata so the origin and edit history can be verified. Adobe, Microsoft, and major camera manufacturers are signatories.
3. In the 2023 Slovakia election audio case, investigators identified the audio as potentially synthetic partly because of:
Correct. AFP Fact Check and DFRLab used a combination of acoustic analysis (prosody, background noise) and linguistic analysis (phrasing atypical of the subject's documented speaking style) β€” a contextual approach rather than relying on a single detection tool.
Not quite. Investigators identified multiple contextual signals: unnatural speech prosody, background noise inconsistencies, and phrasing that didn't match the subject's documented speaking patterns. This illustrates why contextual analysis matters as much as technical detection.
Lab Β· Lesson 3

AI Content Detection Analysis

Practice the contextual verification approach for suspected synthetic media.

Scenario: Suspicious viral video

A 40-second video is going viral on social media showing what appears to be a prominent politician making an inflammatory statement at a press conference. No major news outlets have reported on the event. The account that posted it was created three months ago and has 12,000 followers. Ask your AI lab partner how to approach this investigation β€” covering detection tools, contextual red flags, and how to document your findings if you suspect it's synthetic.

Suggested opener: "I have a suspicious viral video of a politician β€” walk me through how to investigate whether it might be AI-generated." β€” or ask about specific detection tools, artifact analysis, or verification triangulation.
AI Content Detection Lab
AI Lab Partner
You've got a suspicious video with no corroborating news coverage and an account with a short history. That's already two red flags. Ask me about the detection tools available, what visual and audio artifacts to look for, or how to build a contextual case around whether the event actually happened.
Lesson 4 Β· Module 2

Claim Tracing & Network Analysis

Following the chain from original source to viral claim β€” and mapping the amplification networks that make misinformation spread
When a false claim goes viral, does it matter where it started?

In 2018, the Senate Intelligence Committee released reports documenting the Internet Research Agency's (IRA) social media operations during the 2016 US election cycle. The Oxford Internet Institute's Computational Propaganda Project and the firm New Knowledge analyzed IRA content and found that the operation did not primarily originate viral claims β€” it identified organic grievances and amplified them. One documented technique: IRA accounts would create content on fringe platforms, then amplify it through a network of accounts to push it into mainstream feeds, where authentic users would pick it up and spread it further, erasing the trail back to its origin.

Why Tracing Origins Matters

Understanding where a claim originated reveals whether it emerged from an organized influence campaign, a single bad-faith actor, honest error, or a satire account whose label was stripped. Each of these requires a different response. A claim originating from a known state-sponsored influence operation warrants different public messaging than one that started as genuine misinformation from a confused grandmother β€” even if the claim text is identical.

Claim tracing also helps fact-checkers prioritize. When the same claim appears simultaneously across hundreds of accounts that share behavioral characteristics β€” posting at similar times, using similar language, linking to the same obscure sources β€” this is a signal of coordinated inauthentic behavior, not organic spread.

Tools for Claim and Network Analysis

Search Archive
Google Search Operators
Advanced operators narrow searches to find earliest mentions of a claim. Key operators: site: (restrict to domain), before: and after: (date ranges), "exact phrase" in quotes, inurl: for URL keyword search. Combine to trace when exact claim language first appeared.
Social Analysis
CrowdTangle (archived)
Facebook's now-discontinued researcher tool tracked content spread across Facebook and Instagram. Its shutdown in 2024 was criticized by misinformation researchers. Its data archives remain important reference points; Meta's Content Library has partially replaced it for approved researchers.
Network Viz
Gephi
Open-source network visualization tool used by researchers to map how information flows between accounts and communities. Requires data export from platforms or third-party APIs. Used by the Stanford Internet Observatory and EU DisinfoLab in documented investigations.
Domain Research
WHOIS / DomainTools
WHOIS lookups reveal domain registration dates, registrar, and sometimes owner details. Useful for identifying newly registered domains pushing viral claims β€” a common indicator of astroturfing operations. Many influence operations register domains shortly before deploying them.
Bot Detection
Botometer (OSoMe)
Developed by Indiana University's Observatory on Social Media. Scores Twitter/X accounts on likelihood of automated behavior based on activity patterns, posting frequency, and content. Useful for flagging amplification networks, though sophisticated bots can evade detection.
Claim Archive
ClaimBuster / Duke Reporters' Lab
ClaimBuster automatically identifies check-worthy factual claims in text. Duke Reporters' Lab maintains the global fact-checking index. Together they help researchers identify whether a specific claim has already been fact-checked anywhere in the world.

The Anatomy of a Coordinated Campaign

The Stanford Internet Observatory's 2020 report on the "Plandemic" video documented a classic coordinated spread pattern. The video was uploaded on May 4, 2020. Within hours, hundreds of accounts with no prior connection began posting it simultaneously with nearly identical framing language β€” a behavior inconsistent with organic discovery. The SIO found evidence of pre-coordinated distribution planning in private Facebook groups and Telegram channels before the video's public launch.

Key signatures of coordinated inauthentic behavior: synchronized posting times, shared templated language, accounts created in batches (similar registration dates, similar naming patterns), cross-platform simultaneous deployment, and amplification before organic engagement metrics could plausibly have developed.

Case Study EU DisinfoLab β€” Indian Chronicles Investigation, 2019–2020

EU DisinfoLab documented a 15-year operation running 265 fake media outlets in 65 countries, all traced back to a network of front organizations with links to Indian PR firm Srivastava Group. The investigation used WHOIS records, network analysis, and open-source domain tracing to map the operation. Fake NGO websites created recycled UN reports to make their content appear to have been cited at the UN. The operation had been running since 2005, demonstrating that even long-running sophisticated campaigns leave traceable infrastructure when investigators have the right tools.

Putting It Together: A Verification Decision Tree

When confronted with a viral claim, the full toolkit from this module provides a layered approach. Start broad and narrow: check fact-checkers first (fastest, most efficient), then trace the claim's origin using search operators and date filtering, then assess the source via lateral reading, then verify any images or video using reverse search and geolocation, then examine the amplification network if coordinated spread is suspected.

No single tool is sufficient. The power of the verification toolkit is its redundancy β€” multiple independent methods pointing to the same conclusion is the gold standard. One tool flagging suspicious behavior is a signal; three independent methods converging on the same finding approaches a defensible verdict.

Key Terms
Coordinated Inauthentic BehaviorFacebook/Meta's term for organized efforts to manipulate public discourse using fake accounts or coordinated amplification, where the deception lies in the coordination itself rather than necessarily in the content.
BotometerIndiana University tool that scores social media accounts on likelihood of automated behavior based on activity patterns β€” useful for identifying amplification bots.
AstroturfingCreating the false impression of grassroots support or organic spread for a message that is actually coordinated β€” named for the artificial grass brand.
WHOISDomain registration database that reveals when a website was registered and by whom β€” newly registered domains pushing viral claims are a common indicator of influence operations.
Quiz Β· Lesson 4

Claim Tracing & Network Analysis

Three questions β€” select the best answer for each.
1. According to the Senate Intelligence Committee analysis of the IRA's 2016 operations, how did the Internet Research Agency primarily spread misinformation?
Correct. The IRA's documented strategy was amplification of existing grievances rather than pure fabrication β€” which made their content harder to debunk because it often built on real issues, and harder to trace because authentic users eventually carried it further.
Not quite. Research found the IRA primarily amplified organic grievances through coordinated networks rather than purely fabricating content β€” a more insidious approach because authentic users ultimately spread it, erasing the origin trail.
2. In the EU DisinfoLab "Indian Chronicles" investigation, which technical tool was central to tracing 265 fake media outlets back to their source?
Correct. WHOIS records and network analysis of domain infrastructure allowed DisinfoLab to trace hundreds of apparently independent media outlets back to shared registrants and hosting infrastructure, revealing the operation's true scale.
Not quite. WHOIS domain registration records and network analysis were the core tools β€” they revealed that hundreds of apparently independent outlets shared registrants, hosting infrastructure, and creation dates, exposing the coordinated operation.
3. Which of the following is NOT a typical signature of coordinated inauthentic behavior?
Correct. Competing journalists from established organizations independently investigating and discussing a claim is actually a positive signal of genuine newsworthiness β€” it is the opposite of coordinated inauthentic behavior, which involves hidden coordination between actors concealing their relationship.
Not quite. Multiple competing journalists independently engaging with a claim is a sign of genuine newsworthiness, not coordination. Coordinated inauthentic behavior involves hidden relationships between amplifying accounts β€” not the transparent, competitive journalism of rival outlets.
Lab Β· Lesson 4

Claim Tracing & Network Investigation

Practice tracing viral claims and identifying coordination signals.

Scenario: Sudden viral claim with suspicious amplification

A claim appears on X/Twitter that a major bank is about to fail, sourced to an obscure website registered 6 weeks ago. Within 4 hours it has been posted by 400+ accounts, many of which have posted about financial collapse topics repeatedly this month. Your AI lab partner will help you investigate: how to trace the claim's origin, assess the domain, analyze the amplification network, and determine whether this shows signs of a coordinated operation.

Suggested opener: "How do I trace where this bank-failure claim actually originated?" β€” or ask about WHOIS lookups, Google search operators for date-tracing, Botometer, coordinated inauthentic behavior signals, or how to document your findings.
Network Analysis Lab
AI Lab Partner
A 6-week-old domain, 400 accounts posting in 4 hours, and a financial panic narrative β€” there are several threads worth pulling here. Ask me where to start, how to do a WHOIS lookup, what bot signals to look for, or how to use search operators to find the earliest version of this claim.
Module Test Β· Module 2

Detection and Verification Tools

15 questions β€” score 80% or higher to pass this module.
1. Reverse image search works best for finding the origin of an image when using which specific TinEye feature?
Correct. Sorting by oldest date establishes the image's provenance timeline β€” critical for proving a photo predates the event it's claimed to document.
Sorting by "oldest" is the key feature β€” it establishes when the image first appeared, which can prove it predates the event it's falsely attributed to.
2. Which reverse image search engine is noted for superior performance on images originating from Eastern Europe or Central Asia?
Correct. Yandex's Russian-language index and facial recognition capabilities make it particularly effective for images originating in Cyrillic-language regions.
Yandex Images β€” its Cyrillic-language index and superior facial recognition for Eastern European subjects makes it a preferred choice for images from that region.
3. EXIF metadata in a photograph can include all of the following EXCEPT:
Correct. EXIF data is embedded by the camera and includes technical capture information β€” GPS, timestamp, camera model, settings. Social platform uploader identity is not part of EXIF.
EXIF data is embedded by the camera at capture β€” it includes GPS, timestamp, and camera settings. Uploader identity on social platforms is account metadata, not EXIF data.
4. The Stanford History Education Group study found that professional fact-checkers were more accurate at evaluating websites because they:
Correct. Lateral reading β€” immediately checking what external sources say about a site β€” was the defining behavior that made fact-checkers faster and more accurate than historians or students.
Lateral reading was the key β€” opening new tabs to check what credible third parties said about the source, rather than reading within the site itself.
5. The IFCN Code of Principles requires signatory fact-checking organizations to commit to:
Correct. The IFCN Code covers five areas: nonpartisanship, transparent sourcing, transparent funding disclosure, transparent methodology, and an open corrections policy.
The IFCN Code requires nonpartisanship, transparent sourcing, transparent funding, transparent methodology, and an open corrections policy β€” ethical standards, not output quotas.
6. A PolitiFact "Half True" rating most accurately means:
Correct. "Half True" indicates selective accuracy β€” the stated facts may check out but the overall impression conveyed is misleading because of what's left out.
"Half True" means the claim has accurate elements but omits critical context β€” the facts check out partially but the overall impression is misleading.
7. The Internet Archive's Wayback Machine is most useful for fact-checkers when:
Correct. The Wayback Machine stores historical snapshots β€” when a misinformation publisher deletes content after being called out, the archived version provides a permanent record for documentation.
The Wayback Machine's key value is retrieving deleted content β€” when publishers remove misinformation, the archive often retains a snapshot.
8. Which statement about AI image detection tools is most accurate as of 2024?
Correct. The consensus among the verification community is that AI detectors raise useful flags but generate enough false positives that they cannot serve as conclusive evidence. Contextual verification must accompany any detection output.
AI detectors have meaningful false-positive rates β€” they are triage tools that signal the need for closer investigation, not final arbiters of whether content is AI-generated.
9. The C2PA (Coalition for Content Provenance and Authenticity) addresses synthetic media through:
Correct. C2PA works at the creation stage β€” embedding a cryptographic chain of custody so anyone can verify a file's origin and edit history, rather than trying to detect deception after the fact.
C2PA embeds provenance at creation via cryptographic metadata β€” it's a chain-of-custody approach, not a detection approach.
10. In analyzing the 2023 Slovak election deepfake audio, investigators used what combination of methods?
Correct. AFP and DFRLab combined acoustic signals (unnatural prosody, background noise inconsistencies) with linguistic analysis of phrasing atypical of the subject's documented speaking style β€” multi-method contextual verification.
Investigators combined acoustic analysis with linguistic comparison β€” multiple independent signal types converging on the same conclusion, rather than relying on any single detection tool.
11. The "verification through triangulation" principle holds that:
Correct. Real events at real times and places generate multiple independent records β€” video, photos, witness accounts, and journalism from different angles. A single isolated "perfect" image of a major event with no corroboration is suspicious regardless of what detection tools say.
Triangulation means genuine events leave multiple independent records. One isolated perfect image of a major event, with no other coverage, is a red flag regardless of technical detection results.
12. The Internet Research Agency's 2016 operations, as documented by the Oxford Internet Institute, primarily targeted:
Correct. The IRA operation was primarily an amplification operation β€” identifying real grievances and organic community tensions, then using coordinated accounts to amplify them into mainstream visibility.
The IRA's documented approach was amplification of organic grievances β€” making real social tensions more visible and inflammatory, not fabricating entirely new narratives from scratch.
13. WHOIS domain records are useful in misinformation investigations primarily because they can:
Correct. WHOIS records reveal when a domain was registered and sometimes who owns it β€” influence operations often register domains in batches shortly before deployment, a pattern detectable through WHOIS.
WHOIS records expose registration dates and sometimes owner details β€” influence operations often register domains in batches shortly before use, a traceable pattern.
14. Geolocation of a photograph refers to:
Correct. Geolocation is the manual or semi-manual process of reading visual clues β€” signs, buildings, shadows, vegetation β€” to identify where an image was captured, cross-referenced with mapping tools.
Geolocation uses visual clues within the image itself β€” architecture, road signs, shadow direction, vegetation β€” matched against mapping tools to determine where a photo was taken.
15. Which of the following is the correct order for an efficient verification workflow when encountering a viral claim?
Correct. This order moves from fastest and most efficient (existing fact-checks) to increasingly intensive investigation β€” claim origin tracing, lateral source evaluation, visual verification, and network analysis when coordination is suspected.
The efficient order starts with the fastest check (existing fact-checks) and progresses through claim tracing, source evaluation, visual verification, and network analysis β€” saving the most labor-intensive steps for when they're actually needed.