Families across the United States now face a fraud threat that did not exist at scale even two years ago: criminals cloning a relative’s voice from a short audio clip posted online, then calling to demand ransom in a staged emergency. The FBI issued Alert Number I-120525-PSA warning that scammers are sending manipulated or AI-altered photos and videos drawn from social media to make these “virtual kidnapping” calls more convincing. The Federal Trade Commission published separate consumer guidance stating that a single short audio clip is enough for scammers to replicate a loved one’s voice. Both agencies point to the same low-tech defense: a private family safe word that any caller must provide before money changes hands.
Why a family safe word has become the first line of defense
The mechanics of the scam are straightforward. A caller plays a cloned voice that sounds like a son, daughter, or spouse, claims that person is in danger, and pressures the listener to wire money or buy gift cards immediately. The panic is the product. Criminals count on the few seconds of shock to override rational thinking, and the cloned voice supplies just enough realism to keep victims on the line. The FBI alert describes how altered proof-of-life media raises the emotional stakes even further, because victims may receive a photo or video that appears to show their family member in distress.
A pre-agreed safe word short-circuits this sequence. When a household picks a word or phrase that only its members know, any caller claiming to be a relative can be challenged on the spot. If the caller cannot produce the word, the target knows to hang up and verify independently. The FBI alert lists “establishing a private family code word/safe word to verify identity” as a direct mitigation step. The FTC’s guidance on AI-enhanced family emergencies reinforces the same advice: slow down, ask a question only the real person could answer, and resist the pressure to act before thinking.
The hypothesis behind this recommendation is simple. Households that pre-agree on and routinely practice a safe word should be far less likely to send money during a suspicious call than households that have only heard general warnings about phone scams. General awareness tells people that fraud exists; a safe word gives them a concrete action to take in the moment of highest stress. The distinction matters because the scam’s power comes from speed and emotion, not from technical sophistication. Any pause long enough to demand a code word is usually long enough to break the spell.
FBI and FTC evidence backing the safe-word recommendation
The FBI’s December 2025 public service announcement, Alert Number I-120525-PSA, specifically warns that criminals are pulling photos and videos from social media accounts, altering them with AI tools, and sending the results to victims as fake proof that a family member has been captured. The alert does not treat voice cloning as a future risk; it describes an active pattern already generating victim reports. Among its recommended countermeasures, the alert names the family safe word alongside other steps such as contacting the supposed victim directly and limiting personal information shared on public profiles.
The FTC’s consumer advice page on fake emergencies states that AI voice cloning requires only one short audio clip to produce a convincing imitation. That clip can come from a voicemail greeting, a social media video, or any publicly accessible recording. The agency urges consumers to verify any emergency call by hanging up and reaching the family member through a known number. The safe-word tactic fits into this verification framework as the fastest possible check when seconds feel critical.
These warnings did not emerge in isolation. The FBI’s Internet Crime Complaint Center published an earlier alert, I-05142021-PSA, documenting extortion scams that targeted families who posted about missing persons on social media. That 2021 advisory described criminals exploiting emotional vulnerability and demanding ransom payments. The current wave of voice-clone scams follows the same playbook but adds AI-generated audio and visual media to increase believability. The FTC has also highlighted voice cloning as a policy priority through enforcement actions, research initiatives, and public challenges, signaling that the agency views this technology as a present consumer threat rather than a hypothetical risk.
Gaps in the evidence and what families should do first
Neither the FBI nor the FTC has published data measuring how often the safe-word tactic actually prevents payments. No federal dataset breaks down voice-clone scam incidents by year, dollar losses, or victim demographics for 2024 or 2025. The alerts describe the threat and prescribe countermeasures, but they do not include outcome studies comparing families that use safe words against those that rely on general awareness alone. Without that data, the recommendation rests on logical inference and agency authority rather than on measured results.
The absence of victim-level data also means there are no verified case studies showing exactly how a safe word stopped a payment in progress. News reports have supplied anecdotes, but the primary federal sources focus on describing the scams and listing defensive steps. That leaves families with a practical dilemma: how much time and effort should they invest in a mitigation strategy whose effectiveness has not been quantified, even though it is strongly endorsed by the main consumer-protection agencies?
Given that uncertainty, the safest approach is to treat a family code word as one layer in a broader response plan. The first priority is to talk explicitly about the risk. Many victims of virtual kidnapping scams report that they had heard of phone fraud in general but had never imagined hearing a loved one’s voice crying for help. Families should explain, in plain language, that artificial intelligence can now mimic voices and alter photos, and that scammers are already using these tools to manufacture emergencies.
The second priority is to choose a code word or phrase that is memorable, not obviously connected to the family, and never shared outside the household. It should be something every member can recall under stress, including children and older relatives. Families can practice by staging short drills: one person calls another and asks for the word before continuing the conversation. The goal is to make the question automatic, so that it surfaces even when emotions run high.
Third, households should agree on a verification routine that goes beyond the safe word. That includes hanging up and calling the supposed victim on their usual number, reaching out to other relatives or friends who might be with them, and refusing to move money until at least one independent check has been completed. The FBI and FTC both stress the importance of slowing down and breaking the scammer’s control over the timeline. A code word is most effective when it triggers that broader pause.
Finally, families can reduce their exposure by limiting the amount of personal data and audio they share publicly. While it is unrealistic to scrub every recording from the internet, tightening privacy settings on social media accounts and being selective about which videos are public can make it harder for criminals to harvest clean voice samples and convincing photos. This step aligns with the FBI’s broader advice to restrict sensitive information that could be repurposed in extortion schemes.
There is no single measure that can guarantee safety from AI-enabled fraud. But the convergence of FBI and FTC guidance around a simple, low-cost tactic suggests that a private safe word is worth adopting, even in the absence of formal effectiveness studies. It costs nothing, takes minutes to set up, and can be combined with common-sense verification habits. As synthetic voices and images become cheaper and more convincing, that kind of pre-planned friction may be the difference between a terrifying phone call and a drained bank account.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
