How to Verify Suspicious AI Content

Why the Old Tells Stopped Working

For years, spotting an online scam was relatively straightforward. Deceptive emails and messages were often marked by obvious grammatical errors, awkward phrasing, or poorly cropped images. Generative artificial intelligence has fundamentally changed this landscape. According to the FBI's Internet Crime Complaint Center (IC3), criminals use generative AI to synthesize entirely new content, correct human errors, and translate text, removing the traditional warning signs of fraud [IC3 PSA 2024].

In 2025, the IC3 recorded 22,364 complaints involving artificial intelligence, costing Americans nearly $893 million [FBI Press Release 2026]. Because these tools allow foreign actors to communicate without spelling mistakes and generate highly realistic media, relying on visual intuition alone is no longer sufficient. To protect yourself, you must combine technical verification tools with a methodical analysis of behavioral red flags.

The 5-Minute Verification Checklist

When you encounter a suspicious message, image, or audio clip, do not react immediately. Instead, run through this structured verification process:

1. Check for Provenance Signals: If the media is an image or audio file, use a verification tool to check for embedded digital credentials. For example, you can upload files to openai.com/verify to check for supported OpenAI signals [OpenAI Help Center].
2. Analyze Technical Imperfections: Look closely for physical inconsistencies in images (such as distorted hands, inaccurate shadows, or unrealistic accessories) and listen closely to the tone and word choice in audio [IC3 PSA 2024].
3. Verify the Channel Independently: If a caller claims to represent an organization, hang up. Find the official phone number on your statement or the organization's website and call them back directly [FCC Spoofing Guide].
4. Test with a Family Code Word: For urgent calls from relatives, ask for a pre-established secret phrase to confirm their identity [IC3 PSA 2024].
5. Evaluate the Request: Ask yourself if the contact is demanding immediate payment, a transfer of funds, or a security verification code. These demands are behavioral red flags that no technology can disguise [FTC July 2024].

Understanding Technical Provenance Tools

Technical standards have emerged to help identify the origin of digital media. However, a methodical investigator must understand what these tools can and cannot prove. Two provenance standards currently used by OpenAI are C2PA Content Credentials and SynthID watermarks [OpenAI Help Center].

If a verification tool returns a not_detected result, it does not prove the content is human-made. The metadata may have been stripped, the watermark degraded, or the content generated by another company's AI model [OpenAI API]. Provenance tools are useful clues, but they are not definitive proof of authenticity.

Apply It by Channel

Text and Social Media

Criminals use AI to generate voluminous fictitious social media profiles and embed automated chatbots on fraudulent websites [IC3 PSA 2024]. If you are interacting with an online contact, remember that scammers use these tools to generate messages much faster than a human could, allowing them to reach a wider audience with believable content.

Images and Identification Documents

AI-generated images are frequently used to create fake driver's licenses, professional credentials, or profile photos [IC3 PSA 2024]. Zoom in on the details. Look for unrealistic teeth or eyes, indistinct or irregular faces, or unrealistic accessories such as glasses or jewelry [IC3 PSA 2024].

Audio and Voice Cloning

In a typical AI voice clone scam, a fraudster uses a short clip of a loved one's voice to mimic them in an emergency situation [IC3 PSA 2024]. Listen closely to the tone and word choice, which may sound unnatural. If you receive an urgent call, hang up and call your relative back on the number you have saved in your contacts [IC3 PSA 2024].

Video and Real-Time Chats

Deepfake videos are increasingly used to impersonate corporate executives or public figures [IC3 PSA 2024]. In real-time video chats, look for subtle imperfections like lag time or unrealistic movements.

Red Flags That Survive Any Technology

While generative AI can make fake media look and sound real, it cannot alter the fundamental mechanics of a financial scam. The Federal Trade Commission (FTC) and the FBI emphasize that the most reliable indicators of fraud are behavioral:

• The Pressure to Act Quickly: Scammers rely on high-pressure tactics to prevent you from thinking clearly. The FBI urges consumers to "Take a Beat" and assess the situation before acting [FBI Press Release 2026].
• Demands to Move Money: If a caller claims there is fraud on your account and instructs you to transfer funds to "protect" them, it is always a scam [FTC July 2024]. Legitimate banks will never ask you to move your money to a new account or cryptocurrency wallet to keep it safe.
• Requests for Verification Codes: Two-factor authentication codes are security keys meant only for you. Anyone who contacts you and asks for a verification code is a scammer, regardless of how convincing their voice or caller ID appears [FTC March 2024].

Mistakes That Keep People on the Hook

Many victims fall into traps because they rely on outdated assumptions about digital security. Avoid these common errors:

• Trusting Caller ID: Scammers routinely use "neighbor spoofing" or mimic the official numbers of banks and government agencies [FCC Spoofing Guide]. A familiar name on your screen is not proof of who is calling. See our guide on bank impersonation scams for more details.
• Believing Video Proof: Do not assume a contact is real simply because they sent a video or participated in a brief video call. Real-time deepfakes are actively used to bypass visual verification [IC3 PSA 2024].
• Assuming Your Bank Will Recover the Funds: Unlike credit cards, bank transfers and cryptocurrency payments have fewer consumer protections. If you are tricked into voluntarily moving your money, you are highly unlikely to get it back [FTC July 2024].

What to Do If You Shared Information or Paid

If you realize you have interacted with an AI-enabled scammer, take immediate containment steps:

1. Contact Your Financial Institution: If you shared account details, transferred money, or provided a verification code, notify your bank or investment firm immediately to freeze your accounts [FTC July 2024].
2. Report to the FBI: File a detailed complaint with the IC3 at www.ic3.gov. Include names, phone numbers, transaction dates, account numbers, and receiving cryptocurrency addresses [IC3 PSA 2024].
3. Report to the FTC: Submit a report at ReportFraud.ftc.gov to help federal authorities track and shut down imposter networks [FTC June 2026].

Quick Answers

Can a verification tool tell me with 100% certainty if an image is AI-generated?

No. Provenance tools can confirm if a file contains specific, trusted AI signatures (like C2PA or SynthID), but they cannot prove a file is human-made if no signal is detected. Metadata can be stripped, and tools do not detect models from every company [OpenAI API].

Why would a scammer ask for my verification code?

A verification code acts as a second security lock on your account. Scammers who already have your password need this code to bypass two-factor authentication, log in, and drain your funds [FTC March 2024].

Is caller ID spoofing illegal?

Under the Truth in Caller ID Act, it is illegal to transmit misleading caller ID information with the intent to defraud or cause harm. However, spoofing is permitted for legitimate purposes, such as a doctor displaying their office number while calling from a personal phone [FCC Spoofing Guide].