Guide for everyone, updated 11 November 2026
How to protect yourself from AI forgeries, deepfakes and backdatingA plain-English 2026 guide for the US, UK, Europe, Australia and Canada
TimeBinder adds a layer of protection against AI and human fraud and disputes.
One tool among many. This guide covers free and paid options from other organisations too. A TimeBind shows that an exact file existed no later than a recorded time and still matches. It does not detect AI. Your browser makes the fingerprint and the file stays on your device.
Short answer: how do I protect myself from AI forgeries?
You protect yourself from AI forgeries by relying on process, not your eyes and ears. Agree a family safe word. Hang up and call back on a number you already have before you act on any urgent request. Never pay a blackmailer. Keep original files and date them when you create them. Lock down your accounts with a password manager, two-step login and, where you can, a credit freeze.
No tool can tell you for certain whether a photo, video, voice or document is fake. Detectors give a score, not proof. This guide shows what works, which free tools are real, and exactly where to report in the US, UK, Europe, Australia and Canada.
Some links to other tools on this page are affiliate links, marked “affiliate link”. If you buy through one, TimeBinder may earn a commission, at no extra cost to you. Tools are chosen on merit, with caveats. See our affiliate disclosure.
In this guide
New: a free interactive mini-course, Spot AI fakes in 5 short lessons, covers labels, reverse search, voice-clone scam checks, unreliable video tells and dated originals. No checker can reliably identify synthetic media.
- 1. Voice-clone scam callsHow do AI voice-clone scam calls work, and how do I stop them?
- Verify a viral video in 5 stepsA free mini-course: claim, earliest upload, context, labels and reports, and a dated record. No checker can reliably identify synthetic media.
- Check a suspicious voice message in 5 stepsA free mini-course: the ask, a pause, a second channel, saved recordings, and a dated record. No checker can reliably identify synthetic audio.
- Verify a screenshot or fake document in 5 stepsA free mini-course: the claim and source, a pause, the original, file clues, and a dated record. No checker can reliably tell a genuine screenshot from a fake.
- Verify an AI-generated photo of an event in 5 stepsA free mini-course: the claim and source, a pause, the original and other evidence, clues, and a dated record. No checker can reliably tell a real photo from an AI-generated one.
- Verify a viral news clip in 5 stepsA free mini-course: the claim and source, a pause, the outlet and original, clues, and a dated record. No checker can reliably tell a real clip from an AI-generated one.
- Verify a leaked-looking chat or email screenshot in 5 stepsA free mini-course: the claim and source, a pause, the original and the people involved, clues, and a dated record. No checker can reliably tell a genuine screenshot from an edited one.
- Verify a fake product review or testimonial in 5 stepsA free mini-course: the claim and source, a pause, the reviewer and independent sources, clues, and a dated record. No checker can reliably tell a genuine review from a fake or machine-written one.
- Verify an AI-generated receipt or invoice in 5 stepsA free mini-course: name the claim and the payee, a pause, confirm with the issuer and your own records, clues, and a decision with notes. No checker can reliably tell a genuine receipt from a generated one.
- Check a suspicious profile photo or dating profile in 5 stepsA free mini-course: the claim and what is asked of you, a pause, checking outside the picture, clues, and a dated record. No checker can reliably tell a real profile photo from a generated one.
- Verify an AI-edited rental or property listing photo in 5 stepsA free mini-course: the listing claim and what you are asked to pay, a pause, confirming the place with the platform or agent and independent views, clues, and a decision with notes. No checker can reliably tell an AI-edited listing photo from an honest one.
- Check a suspicious celebrity or expert video ad in 5 stepsA free mini-course: the claim and the offer, a pause, checking the seller and the offer outside the ad, clues, and a decision with notes. No checker can reliably tell a real video ad from a deepfake.
- Check a photo, video or audio file offered as evidence in 5 stepsA free mini-course: name the claim, ask for the original file, look for support outside the file, weigh clues both ways, then document. No checker can reliably tell an AI-made file from a real one.
- Check a suspicious video call or remote job interview in 5 stepsA free mini-course: the claim and what you are asked to do, a pause, confirming the person through a channel you already trust, clues, and a decision with notes. No checker can reliably tell a deepfake video call from a real one.
- Check a suspicious website or login page in 5 stepsA free mini-course: name what the page asks, leave by another route, read the whole address, weigh clues both ways, then document. No tool can reliably tell a clone from the real site.
- Check a map or satellite image that may be AI-made or edited in 5 stepsA free mini-course: name the claim, find the provider original, compare with other imagery and sources, weigh clues both ways, then document. No tool can reliably tell AI from real.
- 2. Deepfake video and photosHow can I tell if a video, photo or voice note is a deepfake?
- 3. Sextortion and fake nudesWhat do I do if someone threatens me with fake nude images (sextortion)?
- 4. Fake documents, screenshots and chatsHow do I protect myself from fake documents, receipts, screenshots and chat logs?
- 5. Backdating in disputesWhat is backdating, and how do I protect myself in a dispute?
- 6. Fake IDs and identity theftHow do I protect myself from fake IDs and identity theft?
- 7. AI phishingHow do I spot AI phishing emails, texts and fake websites?
- 8. Protect your own work and recordsHow do I protect my own work and records, and date my files when I create them?
- 9. What to do if it happens to youWhat should I do if I am hit by an AI scam, deepfake or forgery?
- 10. Verified toolsWhich tools can I trust, and which are free?
- 11. FAQFrequently asked questions
What are the quickest ways to protect myself right now?
Do these eight things first. They take less than an hour and stop most AI-assisted scams.
- Agree a family safe word this week, in person. Do not text it or post it.
- Make a rule: for any urgent request for money, codes or secrets, hang up and call back on a saved number.
- Turn on two-step login for email, banking and your phone account, and use a password manager.
- Check your email addresses at Have I Been Pwned and change any reused passwords.
- Freeze your credit where that is available to you (free in the US). See the country table in section 6.
- Keep original photos, files and receipts. Work on copies. Date the original when you create it (section 8).
- Limit how much of your voice and video is public. Make social accounts private where you can.
- Save the reporting links for your country (section 9) in your phone's notes now, not mid-crisis.
1. How do AI voice-clone scam calls work, and how do I stop them?
Scammers copy a voice from short audio clips, then phone you as a relative in trouble or as your boss asking for an urgent payment. The fix is the same every time: hang up, then call the person back on a number you already have, and use a family safe word. Never send money, gift cards, crypto or codes on the strength of a call, voice note or video alone.
The FBI warns that criminals generate short audio clips of a loved one's voice to pose as a relative in a crisis and ask for immediate money, and use AI video in real-time calls to pose as executives or officials (FBI IC3 alert, 3 Dec 2024). The US Federal Trade Commission has warned about the same family-emergency pattern. The Canadian Anti-Fraud Centre describes emergency and grandparent frauds, and Australia's Scamwatch explains how scammers use technology and AI.
What do these scams usually sound like?
| Script | What you hear | What the scammer wants |
|---|---|---|
| Family emergency | A panicked voice that sounds like your child, grandchild or partner. An accident, an arrest, a hospital bill, "I lost my phone", sometimes a "lawyer" or "police officer" who takes over the call. | Money now, often by bank transfer, crypto, gift cards or a courier collecting cash. You are told not to tell anyone (a "gag order"). |
| Boss or supplier payment | A voice or video call, or voice note, from your manager, director or a supplier. "I am in a meeting. Pay this invoice today. Keep it between us." | A payment to a new bank account, a change of bank details, gift cards or access codes. |
Red flags that should stop you
- Urgency and secrecy. "Do it now, don't tell anyone." Real emergencies survive a two-minute call-back.
- An unusual way to pay. Gift cards, crypto, wire transfers, cash pick-up or a courier. The Canadian Anti-Fraud Centre notes that Canadian courts do not let anyone be bailed out with cash or cryptocurrency.
- A new number or a "broken phone" story from someone claiming to be family.
- A call that moves you somewhere else. A "lawyer", "officer" or "bank" takes over and tells you not to call anyone.
- Caller ID that looks right. Numbers can be spoofed, so a familiar name or number proves nothing.
- The voice is good but the questions are not. The caller cannot answer something only that person would know.
How do I set up a family safe word?
The FBI's advice is to create a secret word or phrase with your family to verify identity. Make it work in real life:
- Choose it together, in person. Pick two or three unrelated words. Avoid pet names, birthdays, street names or anything on social media.
- Never put it in writing online. Do not text it, email it or save it in a shared note that others could read.
- Include everyone who might be called, especially older relatives and teenagers. Practise it once so it feels normal.
- Agree what happens if the caller cannot say it: you hang up and call back on a saved number. No exceptions, even if the voice is crying.
- Change it after it has been used or if you think it leaked, and keep a backup question that only the real person could answer.
What should I do during a suspicious call?
- Stay calm and slow it down. Say you will call back. Scammers need you to stay on the line.
- Hang up. Do not press buttons or call the number they give you.
- Call the real person on a number you already have, or ask another family member to reach them.
- Ask for the safe word. If the caller cannot give it, treat the call as a scam.
- If they say they are police, a lawyer or your bank, hang up and call that organisation using a number from its official website or the back of your card.
- Tell someone. Scams rely on you acting alone.
You can also use the free AI Voice Scam Red-Flags Checker: tick what happened on the call or video call and get a plain risk level, next steps and a safe-word helper. It runs in your browser and does not upload the call.
How do I protect my workplace from boss and supplier payment fraud?
- Two-person rule. No new payee, changed bank detail or urgent payment without a second person approving.
- Call back on a known number, never one from the email, voice note or video request.
- Treat video as unproven. A face on a video call is no longer proof of identity. Ask a question only the real person would know, or move the request to a second channel.
- Tell finance staff the script. Scamwatch covers business email compromise, a close cousin of the fake-boss payment request.
2. How can I tell if a video, photo or voice note is a deepfake?
You often cannot tell by looking or listening, and detection tools can be wrong in both directions. Check the source, the context and any provenance data, run a reverse image search, and confirm with the person through a second channel. Treat a detector result as one clue, never as proof.
If you want to test your eye on side-by-side photos, try the free Real or AI? photo quiz. It is a practice test with a real photo next to an AI image we generated. It does not tell you whether any other file is AI-made.
Two further practice quizzes sit with that one: the Real or AI? interiors quiz on listing-style room photographs, and the Real or AI? food ads quiz on menu-style food photographs. Spotting by eye is unreliable. Neither quiz can tell you whether a file you hold is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Two more practice quizzes cover holiday-style photographs and marketplace-style object photographs. Spotting by eye is unreliable. Neither quiz can tell you whether a file you hold is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Two further practice quizzes cover wildlife photographs and pet photographs. Spotting by eye is unreliable. Neither quiz can tell you whether a file you hold is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Three more practice quizzes cover paintings, landscape weather photographs and building exteriors. Spotting by eye is unreliable. None of these quizzes can tell you whether a file you hold is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Two further practice quizzes cover parked-car photographs and night-sky photographs. Spotting by eye is unreliable. Neither quiz can tell you whether a file you hold is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Two more practice quizzes cover plant and flower photographs and city-skyline photographs. Spotting by eye is unreliable. Neither quiz can tell you whether a file you hold is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Two further practice quizzes cover portraits of real people against AI faces and old black-and-white photographs. Spotting by eye is unreliable. Neither quiz can tell you whether a profile photo or an old photograph is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Three more practice quizzes cover railway photographs, odd animals and signs and lettering. Spotting by eye is unreliable. None of them can tell you whether a photo you found online is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Two more practice quizzes cover instrument photographs and mushroom photographs. Spotting by eye is unreliable. Neither can tell you whether a photo you found online is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Two more practice quizzes cover treehouse photographs and turf and earth house photographs. Spotting by eye is unreliable. Neither can tell you whether a photo you found online is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Two more practice quizzes cover miniature and scale-model photographs and swimming pool photographs. Spotting by eye is unreliable. Neither can tell you whether a photo you found online is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Two more practice quizzes cover volcano and lava pictures and giant pumpkin and vegetable pictures. Spotting by eye is unreliable. Neither can tell you whether a photo you found online is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Another practice quiz covers blown glass and glass ornament photographs. Spotting by eye is unreliable. It cannot tell you whether a photo you found online is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Two more practice quizzes cover underwater photographs and cake photographs. Spotting by eye is unreliable. Neither can tell you whether a photo you found online is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Two more practice quizzes cover photographs of abandoned places and crystal and mineral photographs. Spotting by eye is unreliable. Neither can tell you whether a photo you found online is AI-made. A TimeBind shows only that an exact file existed no later than a recorded time and that a copy you hold still matches it.
Older advice said to look for odd hands, teeth or shadows. The FBI still lists signs like distorted hands or feet, irregular faces, inaccurate shadows and lag in video calls, but good fakes have none of them. Use these checks in order.
- Ask where it came from. Who first posted it, and is there an original source you can reach? A shocking clip with no source is a warning sign.
- Check the context. Does it match what the person would really say or do? Does a trusted news source report the same thing?
- Run a reverse image search. Google Lens and TinEye can show where an image appeared before, or whether it is a recycled photo.
- Look for Content Credentials. Upload the file to the free Content Credentials verify tool. If the file carries signed provenance data, it can show who made it and how it was edited. Most files carry none, and a missing credential does not mean a file is fake.
- Check metadata with ExifTool or an online viewer such as exif.tools. Metadata can be edited or stripped, so it is a clue only.
- Try a detector, if you must. See the tools table in section 10. Remember that each gives a probability, and each can miss a fake or flag a real file.
- Verify with a human. Call the person on a saved number. For something that matters, ask for a fresh photo or video that includes a detail only they could supply right now.
Can a detector prove something is AI-made?
No. Detectors estimate a likelihood. Google's Gemini app can check for SynthID watermarks, but it can only recognise content made by Google's own AI tools, and a "not detected" answer does not mean a file is real (Google's help page). The same limit applies to every detector: a miss does not make a file genuine, and a hit does not make it fake.
3. What do I do if someone threatens me with fake nude images (sextortion)?
Do not pay, and stop contact. Block the person, keep the evidence, and report it. Ask each platform to take the images down, use StopNCII.org (adults) or Take It Down (images of anyone under 18 at the time), and report to the police and your national centre. It is not your fault, and a fake image of you is still abuse.
Sextortion is when someone threatens to share an intimate image, real, edited or AI-made, unless you pay or send more. The US Federal Trade Commission notes that these images may be real, digitally altered, or deepfakes created with artificial intelligence, and that the victims can be adults or minors.
What should I do first?
- Do not pay and do not send anything more. Paying rarely ends the threats. The FTC and Australia's eSafety Commissioner both advise against it.
- Cut off contact. Block the person and stop replying. Do not give access to your accounts or devices.
- Keep evidence. Screenshot threats, usernames, profile links, payment demands and dates. Do not delete the conversation. If the image shows someone under 18, do not copy or forward it; report it instead.
- Ask the platform to remove it. In the US, the Take It Down Act requires covered platforms to remove reported intimate images, including AI-made ones, within 48 hours of a request. If a platform does not, report it at TakeItDown.ftc.gov.
- Use the hash tools to stop it spreading. Adults: StopNCII.org makes a digital fingerprint of the image on your own device and shares that with participating companies, so they can detect copies. Images of someone under 18 at the time: NCMEC's Take It Down.
- Report to the police and your national centre. Use the table below.
- Talk to someone you trust. Many people feel shame. You are the victim of a crime.
| Country | Report and get help |
|---|---|
| United States | TakeItDown.ftc.gov (platform failed to remove), FBI tips.fbi.gov or 1-800-CALL-FBI, NCMEC CyberTipline for anyone under 18, CCRI Safety Center, your local police. |
| United Kingdom | Revenge Porn Helpline (adults), Report Harmful Content, Report Fraud if money was demanded, your local police (999 if you are in danger). |
| Europe | Your national police, and Europol's report-a-crime page for where to report in your country. StopNCII.org and Take It Down can be used from anywhere, though participating platforms vary. |
| Australia | eSafety: report image-based abuse and dealing with sexual extortion, ReportCyber, your local police. |
| Canada | Cybertip.ca (reports about children and teens), NeedHelpNow.ca (under 18), report cybercrime and fraud to the Canadian Anti-Fraud Centre, your local police. |
What if it is a child or teenager?
- Stay calm and do not blame them. Fear of getting in trouble is what keeps children silent.
- Use NCMEC's Take It Down or Cybertip.ca in Canada, and report to the police.
- Do not copy, forward or screenshot the image itself. Keep threats and account details instead.
4. How do I protect myself from fake documents, receipts, screenshots and chat logs?
Assume any screenshot, PDF, receipt or chat export can be made or edited in minutes. If one is used against you, ask for the original from the source system, do not argue from memory, and keep your own originals safe. If you may need to rely on your own records, keep the originals and make a dated, checkable record of them early.
AI tools can now draw a convincing invoice, fabricate a conversation in the style of a real chat app, or edit a photo of a receipt. A screenshot only shows pixels. It does not show that the conversation ever happened. See also: can screenshots be used as evidence?
What if someone shows me a document or screenshot that looks fake?
- Do not rely on memory or argue about it. Ask for the original file, not a photo or screenshot of it.
- Ask for the source. For a chat, the message in the app itself or the app's own export. For an invoice, the PDF as sent. For a receipt, the shop's own record.
- Check it with the sender through a known channel. Call the company on its official number and ask them to confirm.
- Look for inconsistencies: fonts that change, rounding errors, wrong tax or business numbers, impossible dates, addresses that do not match, a profile picture or name format that does not match the real app.
- Check metadata, but only as a clue. Details such as the app that created the file, or software that did not exist on the claimed date, can reveal a problem. Clean metadata proves nothing, because it can be stripped.
- Do not delete anything. Preserve your own messages, emails and files from that time.
- For anything serious (court, a regulator, a large sum), speak to a lawyer. Courts and experts decide what is authentic, not a website.
Red flags in a document or screenshot
- It arrives only as an image or screenshot, and the sender will not give the original.
- The sender pushes you to pay or agree immediately.
- Bank details changed from previous invoices.
- The date, time zone, phone number format or app layout looks slightly wrong.
- Parts of the text are sharper or blurrier than the rest.
- The file's metadata date does not fit the story, for example created after the date it claims.
How do I make my own screenshots and exports more reliable?
- Capture the whole screen including the date, time, contact name and web address, not a tight crop.
- Keep the original export from the app (chat export, email as .eml or .pdf), not only a screenshot.
- Save it the same day and do not edit, crop or re-save the original.
- Make a dated, checkable record straight away. A TimeBind can show that this exact file existed no later than a recorded time, and the free file comparison checker shows whether two files are identical.
5. What is backdating, and how do I protect myself in a dispute?
Backdating is making a document, message, photo or file look as if it was created earlier than it really was. File dates and metadata are easy to change, and AI makes convincing old-looking papers easier to produce. Protect yourself by keeping original records made at the time, putting agreements in writing the same day, and dating your files when you create them.
Most disputes come down to "what was agreed, and when?". The strongest position is simple: you have records made at the time, kept in their original form, with a date that does not rely only on you or on a setting someone can change. The table shows what to keep in common situations. It is general information, not legal advice.
| Situation | What to keep and date at the time | Backdating red flags |
|---|---|---|
| Landlord and tenant | Signed lease and every amendment. Move-in condition report and photos (with timestamps) on day one. Rent receipts. Repair requests sent by email or the portal. | A "condition report" or lease addendum nobody saw before. A repair notice that appears only after a dispute starts. |
| Contractors and builders | Written quote, signed scope and every variation, confirmed in writing the same day. Dated progress photos. Invoices as originally sent. | A "signed" variation that you never saw. An invoice whose metadata date is earlier than the job. |
| Workplace | Your contract and every version you were given. Dated notes of meetings (who, what, when). Performance reviews and warnings you received, as originals. | A warning or policy that is "dated" before the event but first appeared later. Documents that exist only as screenshots. |
| Insurance | Photos or video of valuables, serial numbers, receipts and valuations before any loss. Policy documents. All claim correspondence. | Receipts or photos created after the loss but dated before it. Edited images of damage. |
| Custody and family disputes | Message exports, handover and schedule logs, and records of agreed changes. Use any communication tool a court or agreement requires. Take legal advice early. | Edited chat screenshots. Recordings or messages that do not match the originals. |
| Creative work | Draft and project files, raw photos, stems, sketches. A dated record of the finished work before you share or send it. | A "first version" with a file date that cannot be matched to any earlier draft or backup. |
How can I date a file so it cannot be quietly backdated?
| Method | What it shows | Where it falls short |
|---|---|---|
| File date and metadata | A date your device or app recorded. | Can be changed in seconds, and copying or syncing can alter it. |
| Email to yourself or a cloud upload date | A message or file existed in an account at some time. | Depends on one company's records. You may not be able to prove it later. |
| Notary or witness | A person saw or certified something on a day. | Valuable for signing, but it does not fingerprint a digital file you can compare later. |
| Wayback Machine (web pages) | An archived copy of a public web page. | Only for public pages. Not for private files. |
| OpenTimestamps | A free, open standard that anchors a file's fingerprint to the Bitcoin blockchain. | Technical to use and to verify. You must keep the .ots proof file safe. |
| TimeBinder | A dated, checkable record that an exact file existed no later than a recorded time and still matches. | Does not show who made the file, who owns it, or that its contents are true. |
If you need the full background, see how to prove a file existed before a date, hash timestamp vs notarisation and Wayback Machine as proof.
6. How do I protect myself from fake IDs and identity theft?
Make it hard for criminals to use your identity: freeze your credit where possible, use a password manager and two-step login, check whether your email appears in breaches, and stop sending unmarked ID photos. If your identity is misused, report it quickly using the country table below.
AI can produce realistic fake IDs and doctored photos of real ones. The FBI lists fake driver's licences and credentials among the uses of AI-generated images in fraud (IC3 alert). The best protection is to make your real details useless to someone else.
What are the essential steps?
- Use a password manager so every account has a unique password: Bitwarden, 1Password, Proton Pass (affiliate link) or KeePassXC. The UK's NCSC explains password managers.
- Turn on two-step verification on email, banking, phone account and social media. Prefer an authenticator app, a passkey (FIDO Alliance) or a security key over text-message codes. See CISA's More than a Password and the NCSC's 2-step verification guide.
- Check for breaches at Have I Been Pwned. If your email appears, change that password everywhere you used it.
- Freeze your credit where you can (see the table).
- Be careful sending ID. Send an ID copy only when you must. Write "For [name], [purpose], [date] only" across the copy. Never send ID by an unverified link.
- Check credit reports regularly, and set up alerts from your bank.
| Country | Protect and report |
|---|---|
| United States | Freeze your credit for free with Equifax, Experian and TransUnion; the FTC explains credit freezes and fraud alerts. Free reports at AnnualCreditReport.com. Report identity theft at IdentityTheft.gov. |
| United Kingdom | Report to Report Fraud (formerly Action Fraud). Cifas Protective Registration is a paid service that asks organisations to take extra checks; read its page for the current terms. See also Take Five to Stop Fraud. |
| Europe | Report to your national police. Ask your bank and national credit-reference agency about fraud alerts. Rules differ by country, so check your national consumer or data-protection authority. |
| Australia | Get free support from IDCARE (1800 595 160). Ask the credit reporting bodies, such as Equifax Australia and Experian Australia, about a credit ban. Report to ReportCyber. |
| Canada | Report to the Canadian Anti-Fraud Centre at reportcyberandfraud.canada.ca (1-888-495-8501). Ask Equifax Canada and TransUnion Canada about fraud alerts and security freezes. |
What if my identity has already been misused?
- Contact your bank and any affected company and ask them to close or freeze the accounts.
- Place a freeze or fraud alert, then report it using the table.
- Change passwords on email first, then banking and everything else, and turn on two-step login.
- Keep a log: dates, names, reference numbers and copies of what you sent.
- Replace stolen or copied ID documents as advised by the issuing agency.
7. How do I spot AI phishing emails, texts and fake websites?
Do not rely on spotting bad spelling, because AI writes flawless messages. Instead, never use the link or number in the message. Go to the website or app yourself, or call the organisation on a number you already trust. Then report the message.
AI lets scammers write fluent, personal messages in any language. The FBI notes criminals use AI text to reduce the spelling and grammar errors that once gave scams away (IC3 alert). So change the habit: judge by what the message asks you to do.
Red flags that matter more than spelling
- Urgency or fear: your account will be closed, a parcel is held, you owe tax or a fine.
- A request to log in, pay, verify or download from a link, attachment or QR code you did not ask for. The FTC has warned about QR codes placed in public places.
- A sender address or web address that is almost right, such as an extra letter or a different ending.
- A message that asks for a code (a one-time code is a key to your account, and no genuine company asks for it).
- An unexpected voice, text or video follow-up that pushes you to act on the email.
- Stop and do not click. Hover to check the web address, or better, do not use the link at all.
- Go to the real site or app by typing the address or using a bookmark.
- Check using another channel. Ring the company on the number on your card or official website.
- Use a password manager. It will not offer to fill your login on a look-alike site, which is a useful warning.
- Report and delete. See the table below.
| Country | Report phishing here |
|---|---|
| United States | ReportFraud.ftc.gov, and forward phishing emails to APWG at reportphishing@apwg.org. CISA explains how to recognise and report phishing. |
| United Kingdom | Forward emails to report@phishing.gov.uk and texts to 7726 (free), as explained on GOV.UK. See the NCSC's phishing guide. |
| Europe | Report to your bank, your national police or cybercrime unit and the impersonated organisation. Europol's report-a-crime page points to national reporting. |
| Australia | Scamwatch: report a scam and ReportCyber. Read Scamwatch's phishing scams page. |
| Canada | Report cybercrime and fraud or call the Canadian Anti-Fraud Centre on 1-888-495-8501. See Get Cyber Safe. |
8. How do I protect my own work and records, and date my files when I create them?
Keep originals untouched, work on copies, keep backups in more than one place, and date the original when you create it. A dated, checkable record made at the time is far harder to dispute than a file date or a screenshot made later.
The aim is to be able to say, later: "This is the exact file I had, and here is a record made at the time that it existed." That helps in a dispute about photos, documents, designs, music, writing, receipts and agreements.
- Keep the original, and never edit it. Copy the file and work on the copy. Set the original to read-only.
- Keep the metadata. Camera photos hold the date, device and often location. Share a cleaned copy, but keep the original with its metadata. (Strip location before posting publicly.)
- Back up using the 3-2-1 idea: three copies, on two kinds of storage, with one off-site. The NCSC explains backing up your data.
- Name and organise files clearly. A dated folder for each matter, trip, job or project makes it easier to show a sequence of events.
- Make a dated, checkable record when you create or receive something important. Options include OpenTimestamps (free, technical) and TimeBinder (below).
- Keep the proof with the file. Store the certificate or proof file next to the original, and back both up.
- Check later. If two copies are in dispute, compare them. Our free file comparison checker shows whether two files are exactly the same, using fingerprints made in your browser.
How does TimeBinder fit in?
TimeBinder is one tool for step 5. A TimeBind is a dated, checkable record that an exact file existed no later than a recorded time and still matches. Your browser makes a digital fingerprint (a SHA-256 hash) of the file, and the file stays on your device. TimeBinder records the fingerprint with the time and gives you a Proof of Time Certificate. Your first Pooled TimeBind is free with no card, then from US$5, and an ASAP TimeBind is US$15.
Start with the five short answers, for example how to timestamp a document for free and how to tell if a file has been changed, or read how to prove a photo is not AI for the honest limits of photos and video.
9. What should I do if I am hit by an AI scam, deepfake or forgery?
Stop, protect your money and accounts, keep evidence, then report it. The first hour matters most for money: call your bank. Then report to the police and your national fraud centre using the table below.
- Stop contact and do not pay more. Do not argue with the scammer.
- Call your bank or card provider immediately if money moved, and ask to stop or recall the payment.
- Secure your accounts. Change passwords starting with email, turn on two-step login, and log out other devices.
- Keep evidence. Save messages, numbers, usernames, links, receipts, emails (with headers if you can) and screenshots, and write a short dated timeline. Do not delete the conversation.
- Report it using the table below. Reports help others and can help recover money.
- Freeze credit and report identity theft if personal details were used.
- Tell the people affected (family, your employer, your contacts) so they do not fall for the next step.
- Get support. Scams are upsetting. Free victim support exists in most countries.
| Country | Fraud and scams | Online abuse, deepfakes and images | Identity theft and support |
|---|---|---|---|
| United States | ReportFraud.ftc.gov (FTC), IC3.gov (FBI Internet Crime Complaint Center), your local police. FTC: what to do if you were scammed. | TakeItDown.ftc.gov, StopNCII.org, NCMEC Take It Down, CyberTipline, tips.fbi.gov. | IdentityTheft.gov; VictimConnect. |
| United Kingdom | Report Fraud (formerly Action Fraud), 0300 123 2040; Scotland: Police Scotland. Phishing: report@phishing.gov.uk and 7726. | Revenge Porn Helpline, StopNCII.org, Report Harmful Content. | Victim Support; Cifas identity advice. |
| Europe | Your national police and bank. Start at Europol's report-a-crime page to find where to report in your country. | StopNCII.org, Take It Down and your national police and online-safety helpline. | Your national data-protection or consumer authority and national victim support. |
| Australia | Scamwatch: report a scam, ReportCyber, your bank, your local police. | eSafety: report image-based abuse. | IDCARE, 1800 595 160. |
| Canada | Report cybercrime and fraud / Canadian Anti-Fraud Centre, 1-888-495-8501; your local police. | Cybertip.ca, NeedHelpNow.ca, StopNCII.org. | Equifax Canada and TransUnion Canada for fraud alerts; local victim services. |
Numbers and rules change. In an emergency, call your local emergency number first (911 in the US and Canada, 999 in the UK, 112 in the EU, 000 in Australia).
10. Which tools can I trust, and which are free?
The tools below are real, working services we checked on 11 November 2026. Use reporting services and provenance data first, then metadata and reverse image search, and treat AI detectors as a weak clue. Most online checkers upload your file to a company, so never use them for intimate, children’s or confidential files.
Tools, prices and free tiers change. Where we say "uploads the file", assume the company can see it. TimeBinder and the file comparison checker make the fingerprint in your browser, and the file stays on your device.
We also have three free interactive pieces that stay in your browser: the Real or AI? photo quiz, the viral AI fakes quiz (each answer links to the reporting) and the AI Voice Scam Red-Flags Checker. They are for practice and awareness. None of them detects whether a file, photo, voice or call is AI-made.
Two new free tools sit alongside those quizzes: the viral product-video red-flags checker for clips that look like a real product, and the AI watermark and label decoder for SynthID results, C2PA manifests and platform AI labels.
Two further free checklists sit with those tools: the AI agent file-access checklist before you give an assistant write access to work files, and the AI visualization handoff checker before you send a generated or assisted view. Neither one identifies generated media, and the file-access checklist is not a backup.
Three further free checklists sit with those tools: the chatbot article-reuse checker when an answer looks close to a published article, the photo contest entry checklist before you submit a photo or video, and the marketplace listing-photo checklist before you publish a product listing. None of them identifies generated media, and the article-reuse checklist is not legal advice.
Three more free checklists sit with those tools: the unexpected track on your artist profile checklist when a release you did not send appears on an official profile, the AI CAD add-in checklist before you install an assistant beside CAD files, and the assignment AI-use checklist before you submit an assessment or a paper. None of them identifies generated media, and the CAD checklist is not legal advice.
| Tool | Use it for | Cost | Heads-up |
|---|---|---|---|
| Reporting and help | |||
| ReportFraud.ftc.gov | Report scams and fraud to the US FTC | Free | US only. |
| StopNCII.org | Stop intimate images spreading (adults) | Free | Makes the fingerprint on your device. Adults only. |
| Take It Down (NCMEC) | Images taken when you were under 18 | Free | Run by the US child-protection charity NCMEC. |
| Check where media came from | |||
| Content Credentials verify | Read signed origin and edit history in a file | Free | Most files have no credentials. Missing does not mean fake. |
| Google Lens | Reverse image search | Free | Uploads the image to Google. |
| TinEye | Reverse image search for earlier copies | Free (limits) | Uploads the image. Sometimes shows a browser check first. |
| Metadata and image forensics | |||
| ExifTool | Read all metadata in photos, PDFs and more | Free | Runs on your computer. Technical. |
| exif.tools | View metadata in your browser | Free | Says it runs in your browser. Check its privacy notes. |
| Forensically | Magnifier, error-level and clone detection | Free | Says images stay on your computer. Results need expertise. |
| FotoForensics | Error-level analysis for photos | Free | Public site. Uploads are not private. Do not use for sensitive images. |
| AI and deepfake checkers (a score, not proof) | |||
| Hive Detect (now Deploy) | Estimate if an image, video or audio is AI-made | Free tier | Uploads the file. Can miss fakes or flag real files. |
| AI or Not | Estimate if an image is AI-made | Free and paid | Uploads the file. Can be wrong. |
| Deepfake-o-meter | Run several research detectors on media | Free account | University research tool. Uploads the file. |
| Deepfake Total (Fraunhofer AISEC) | Analyse suspicious audio | Free | Uploads the audio. Research tool. |
| Gemini SynthID check | Check for Google AI watermarks | Free (Google account) | Only finds Google AI content. Uploads the file to Google. |
| Account and identity safety | |||
| Have I Been Pwned | See if your email appears in data breaches | Free | Never enter your password on the email search. |
| Bitwarden | Password manager (open source) | Free and paid | Also stores two-step codes on paid plans. |
| 1Password | Password manager and passkeys | Paid | Free trial may be offered. Check current terms. |
| Proton Pass (affiliate link) | Password manager with email aliases | Free and paid | Affiliate link. Not a record of your files. |
| KeePassXC | Offline password manager | Free | Stores passwords in a file on your device. |
| Credit freezes | Block new credit in your name (US) | Free | See the country table in section 6 for other countries. |
| Dating your files | |||
| OpenTimestamps | Free open standard for Bitcoin timestamps | Free | Technical. Keep the .ots proof file. |
| TimeBinder | A dated, checkable record that an exact file existed no later than a recorded time | First Pooled TimeBind free, no card; then from US$5 | Fingerprint made in your browser; file stays on your device. Does not detect AI or show who made a file. |
| Free file comparison checker | See if two files are exactly the same | Free | Fingerprints made in your browser. Shows same or different, not which is older. |
Frequently asked questions
Can I tell if a voice on the phone is an AI clone?
Not reliably. A cloned voice can sound exactly right, so do not trust the sound alone. Hang up and call the person back on a number you already have, and use a family safe word. No consumer tool can promise to catch a cloned voice on a live call.
What is a family safe word and how do I choose one?
It is a secret word or phrase that only your family knows, used to check that a caller is who they claim to be. Choose two or three unrelated words together in person. Avoid names, dates or anything on social media, and never text or post it. If the caller cannot say it, hang up and call back.
Should I pay if someone threatens to share fake nude images of me?
No. Stop contact, block the person, keep evidence and report it. Paying rarely ends the threats. Ask the platform to remove the image, use StopNCII.org (adults) or Take It Down (images of anyone under 18), and report to the police and your national fraud centre.
Can I get a fake nude of me removed from the internet?
Often, yes. Report it to each platform. In the US, covered platforms must remove reported intimate images, including AI-made ones, within 48 hours of a request, and you can report a platform that does not at TakeItDown.ftc.gov. StopNCII.org and Take It Down help stop copies spreading on participating platforms. No service can promise to remove every copy.
How can I check if a photo or video is real?
Check where it came from, run a reverse image search with Google Lens or TinEye, look for Content Credentials, and check the metadata. A detector can add a clue, but none gives proof. Confirm anything important with the person or source through a second channel.
Do AI deepfake detectors work?
Sometimes. They give a probability, and they can miss a fake or flag a real file, especially after a file has been compressed or re-saved. Treat them as one clue. Do not upload private or intimate images to them.
What is backdating, and why is AI making it worse?
Backdating is making something look older than it is. File dates and metadata are easy to change, and AI can produce convincing documents, receipts and chat screenshots. Records made at the time, kept in original form and dated by a method that does not rely on you, are harder to dispute.
How can I prove a file existed on a certain date?
Keep the original, and make a dated, checkable record when you create it. A TimeBind, or a free OpenTimestamps proof, can show that an exact file existed no later than a recorded time and still matches. Neither shows who made the file or that its contents are true.
What does a TimeBind prove?
It shows that an exact file existed no later than a recorded time and that a copy shown later still matches its fingerprint. The fingerprint is made in your browser and the file stays on your device. It does not show who made the file, who owns it, or whether AI was used, and it is not legal advice.
Is there a free way to date my files?
Yes. Your first Pooled TimeBind on TimeBinder is free with no card, then it is from US$5. OpenTimestamps is also free, but more technical. Cloud history, email to yourself and file metadata are free too, but they depend on records that can be changed or lost.
Where do I report an AI scam in my country?
US: ReportFraud.ftc.gov and IC3.gov. UK: Report Fraud (formerly Action Fraud). Australia: Scamwatch and ReportCyber. Canada: the Canadian Anti-Fraud Centre. Europe: your national police, starting at the Europol report-a-crime page. The table in section 9 has every link.
Should I freeze my credit to prevent identity theft?
In the US, a credit freeze is free and stops new credit being opened in your name, according to the FTC. Other countries have different tools such as fraud alerts, credit bans or protective registration. See the table in section 6.
Does this guide replace legal advice?
No. It is general information. If you face a formal dispute, a court case or a regulator, speak to a qualified lawyer where you live.
Related guides and tools
- Guide for lawyers: preventing AI forgeries and backdatingOur companion guide for law firms and their clients
- Spot AI fakes in 5 short lessonsA free mini-course on image labels, reverse search, voice-clone scams, unreliable video tells, and dated originals. No checker can reliably identify synthetic media.
- Answers: proving what a file was, and whenFive short answers to common file-proof questions
- Free file comparison checkerDrop two files and see if they are identical
- Real or AI? photo quizTen rounds of a real photo against an AI image
- Could you spot these viral AI fakes?Ten questions on recent, widely reported AI fakes
- AI Voice Scam Red-Flags CheckerA checklist for suspicious calls, with next steps
- How to prove a photo is not AIThe honest limits of photos and video
- Can screenshots be used as evidence?Why screenshots are weak on their own
- What to do if the other side says your evidence is a deepfakeBefore a dispute turns on authenticity
- Content Credentials (C2PA) vs a timestampDifferent questions, different tools
- Screenshot timestampingDate your screenshots and exports
- Creators: date your workFor photographers, writers, musicians and designers
About this guide
This is general information, not legal advice. Laws and reporting services differ by country and change over time. Links to other organisations are for your convenience and we do not control them. We checked every link on 11 November 2026. TimeBinder is our product and is listed alongside other options.
Is a TimeBind right for your records? See pricing and the short answers.
Date your important files while they are still yours alone
TimeBinder adds a layer of protection against AI and human fraud and disputes: a dated, checkable record that an exact file existed no later than a recorded time.
- Open the app and sign in with email or Google.
- Add your files. They stay on your device.
- Seal them and download your Proof of Time Certificate.
First Pooled TimeBind free, no card, then from US$5. See pricing.