A signal is not proof
AI detectors, metadata, writing patterns and visual artefacts can point you towards questions. Treat them as leads to investigate, not automatic verdicts about truth, authorship or intent.
LEARN · VERIFY · DECIDE
The Another AI Lie Evidence Centre
Online information rarely arrives with a reliable label. This centre helps you examine what is in front of you, separate meaningful evidence from weak signals and decide what to check next—without pretending that one score, detector or search can provide certainty.
Start with the evidence, not the verdict
A suspicious phrase, an unusual image detail or a high AI-likelihood score can be useful, but none of them proves a conclusion on its own. Reliable verification comes from combining independent clues, understanding their limits and looking for information that could disprove your first impression.
Context changes meaning
A genuine quotation can still mislead when it is clipped, old or attached to the wrong event. Check who created the material, when it appeared and what surrounded it.
Independent sources matter
Ten websites repeating the same unsupported claim are not ten independent confirmations. Trace reports back to their earliest available source and look for separate evidence.
Uncertainty is useful
An honest result may be inconclusive. Knowing that the available evidence is weak is more valuable than receiving a confident answer that the evidence cannot support.
A practical verification method
Use these six steps whenever a claim, image, video or message matters enough to influence what you believe, buy or share.
- Pause before reacting.Urgency, outrage and fear make people more likely to share first and check later. Give yourself enough time to inspect the material.
- Define the exact claim.Write down what is actually being asserted. Separate facts that can be checked from opinion, prediction, humour and emotional framing.
- Identify the original source.Look beyond the account or page that reposted it. Find the earliest accessible publication, full quotation, original image or complete video.
- Check date, place and context.Old or genuine material is often reused with a new caption. Confirm that the time, location and surrounding events match the claim.
- Seek independent corroboration.Use sources with direct access to the evidence. Prefer official records, named experts, primary documents and reputable reporting over repeated summaries.
- Decide in proportion to the evidence.You do not have to choose only “true” or “false”. The responsible conclusion may be likely, doubtful, misleading, unverified or currently unknowable.
Choose the right checks for the material
Different formats leave different kinds of evidence. Start with checks suited to what you are examining.
Text and factual claims
Separate checkable statements from persuasive language. Search distinctive phrases, inspect cited sources, compare dates and look for primary documents that support the central claim.
Images
Look for earlier appearances, alternate crops and the original upload. Check whether lighting, reflections, edges and written text are internally consistent, while remembering that compression and editing can create innocent artefacts.
Video and audio
Find the full-length source and compare key frames with trusted coverage. Watch for abrupt transitions, mismatched lip movement or unnatural sound, but do not treat visual oddities alone as proof of a deepfake.
Email and messages
Verify the sender through a separate trusted route. Inspect the real destination of links, question unexpected attachments and never let a message rush you into disclosing credentials or making a payment.
Reviews and recommendations
Look for specific experience, balanced detail and a believable history across multiple platforms. Repeated wording and sudden bursts of similar ratings can justify further checking but may have several explanations.
Student or professional work
Use drafts, sources, revision history and conversation about the work. A detector score should never be treated as proof of misconduct or substituted for a fair evidence-led process.
How to read an Another AI Lie result
The checker groups different questions instead of collapsing everything into one dramatic verdict. Each result should be read alongside its explanation and the limitations of the available material.
Signal
What it can indicate
What it cannot prove
Evidence status
Whether public sources support, contradict or leave the checkable claim unresolved.
That every statement is true, or that the source is honest in every context.
Evidence coverage
How much of the checkable material was assessed and how many sources were used.
That every possible source was found or that an unresolved claim is false.
AI-writing signals
Limited writing-style patterns such as unusually uniform sentence lengths, formulaic transitions or repeated sentence openings.
Who wrote the material, which tool was used, or whether AI involvement occurred.
Internal repetition
Repeated phrases or wording inside the submitted text.
Plagiarism, copyright infringement or copying from an outside source.
Public-source similarity
Possible overlap with public web material when that check is included in a fuller investigation.
A comprehensive plagiarism search, authorship or deliberate copying.
Important: every signal is a starting point for investigation, not a verdict. AI-writing signals do not establish authorship, and an unresolved evidence status does not mean a claim is false.
Read the full methodology and limitations or view a representative Full Evidence Report.
Red flags worth investigating
Red flags are prompts to slow down. Their presence may have an innocent explanation; their absence does not guarantee reliability.
No identifiable original source or author
A headline that overstates what the evidence says
Pressure to act, pay or share immediately
Dates or locations that do not fit the claim
Statistics with no sample, method or source
Quotations that cannot be found in full
Images shown only as low-quality screenshots
Many reports that all trace to one claim
Absolute certainty about a complex question
Requests to ignore normal verification steps
The evidence-before-sharing checklist
Before passing something on, ask: Can I name the original source? Is the date relevant? Does the full material support the caption? Have independent sources confirmed the central claim? Am I sharing because the evidence is strong, or because the content produced a strong reaction?
If one of those answers is unclear, add context or wait. “I have not verified this” is often the most accurate and responsible conclusion available.
Practical verification guides
Use a focused guide when you need to investigate a particular type of claim or material.
How to fact-check an online claim
Trace a claim to its original source, examine the evidence and seek independent confirmation.
How to check whether an image may be AI-generated
Use provenance, context, reverse-image search and visual clues without overstating certainty.
How to verify a video before sharing
Check the source, date, location, editing context and the precise claim attached to a clip.
How to check a source, citation or expert quote
Trace references to their originals and assess relevance, method, context and independence.
How to recognise scams and impersonation attempts
Identify urgency and authority tactics, then verify surprising requests through trusted channels.
How to read AI-detection results responsibly
Understand what detector signals can suggest, what they cannot prove and what to investigate next.
How to verify claims on a CV
Separate plausibility, public support and evidence-holder verification without treating missing results as deception.
How to spot a fake product review
Assess specific claims, purchase context, account patterns and independent evidence rather than style alone.
Did AI write this?
Separate language signals, factual support, source overlap and process evidence before reaching a conclusion.
How accurate are AI detectors?
Understand false positives, thresholds and why a score cannot prove authorship or misconduct.
Is this email a scam?
Verify a suspicious request outside the message without clicking its links or trusting its contact details.
How teachers can check suspected AI work fairly
Use policy, sources, drafts and conversation while keeping detector signals in their proper place.
How to verify a screenshot
Recover the original source and cropped context, then test the visible claim separately.
Find the original context of a viral video
Trace the earliest version and verify its date, place, sequence and attached caption.
Have something you want to check?
Use the free checker to organise the first signals, then follow the evidence wherever it leads.