Why Your Family Needs a Code Word in the Age of AI Kidnapping Scams

Comment ‘Blog’ and I’ll send you the link to my site where you can read the full blog with all citations included.

An evidence-based guide for parents raising children in the age of artificial intelligence

The Landscape Has Shifted

When today’s parents were growing up, personal safety education looked different. Children were taught to walk away from strangers offering candy, avoid getting into unfamiliar cars, and never go anywhere without telling a trusted adult. That wisdom still holds — but it is no longer enough.

Predators have evolved. And so must our conversations with children.

Criminals now harvest publicly available information — photos from family Instagram accounts, voice snippets from TikTok videos, even geographic details tagged in social media posts — to construct deeply convincing deception scenarios targeting both parents and children. The tools they use are accessible, affordable, and frighteningly accurate. Understanding this new threat landscape is the starting point for every protective conversation a parent can have.

How AI Has Changed the Threat

Voice Cloning: When a Familiar Voice Becomes a Weapon

Artificial intelligence can now replicate a person’s voice with approximately 95% accuracy using as little as three seconds of recorded audio. That is shorter than the time it takes to say “Hi, it’s me, call me back.” Every video a child posts on social media, every voice note a teenager sends publicly, every reel shared on Instagram — all of it becomes raw material for potential misuse.

Criminals feed these audio samples into widely available AI voice synthesis tools to produce fake recordings of loved ones in distress. The resulting audio is used in two connected ways: to convince parents that their child has been taken (virtual kidnapping scams), and to convince children that a trusted adult has sent someone to pick them up. In one documented case, a Florida mother lost $15,000 after scammers used an AI-cloned voice of her daughter to simulate a kidnapping scenario.

The FBI has issued formal warnings confirming that criminals scan publicly available social media photos, then alter them using AI to fabricate convincing “proof of life” images in fake ransom demands. In 2024, a deepfake incident occurred every five minutes globally, and digital document forgeries surged 244% year-over-year. The same images parents innocently post — school portraits, holiday photos, sports day pictures — become tools for manipulation.

Research from Australia’s eSafety Commissioner found that on one child exploitation image-sharing site containing at least 45 million images, approximately half the material appeared to have been sourced directly from social media platforms, labelled in organised folders by platform name. By the time a child turns 13, an average of over 1,300 photos of them have been posted online by their parents.

AI-Driven Grooming and Behavioural Profiling

Beyond voice cloning, the United Nations has warned that predators are now using AI to analyse a child’s online behaviour, emotional state, and interests in order to tailor grooming strategies with precision. Organisations advocating for children report that technology-facilitated child abuse cases in the United States rose from 4,700 in 2023 to over 67,000 in 2024 — a 1,325% increase in a single year.

UNICEF put it plainly in early 2026: “The harm from deepfake abuse is real and urgent. Children cannot wait for the law to catch up.”

What the Research Says About Teaching Child Safety

Moving Beyond “Stranger Danger”

The phrase “stranger danger” — once the cornerstone of child safety education — has been officially retired by leading child protection organisations. The National Center for Missing and Exploited Children (NCMEC) does not support the stranger-danger model, citing research showing that the majority of missing and exploited children are taken by someone they already know. Children conditioned by the stranger-danger framework often fail to recognise danger from friendly, familiar-seeming individuals precisely because they do not match the mental image of a “stranger”.

Research published in Child Abuse Review found that even children aged 8 to 10 remain vulnerable to approaches by unfamiliar adults — particularly when the approach is friendly rather than threatening. The implication is direct: children need more than a rule. They need a practised, embodied response.

Behavioural Skills Training: The Evidence-Based Approach

The most well-researched method for teaching children to respond to abduction lures is Behavioural Skills Training (BST) — a structured intervention that combines instruction, modelling, rehearsal, and feedback. Multiple peer-reviewed studies confirm that BST is effective in teaching children to refuse lures and remove themselves from dangerous situations.

A landmark study published in the Journal of Applied Behavior Analysis by Johnson et al. (2006) compared BST alone to BST with an added in situ training component — where children are tested in real-world scenarios without knowing they are being observed. Both approaches were effective, but the combination produced stronger generalisation. A 2010 study replicating BST with children diagnosed with autism spectrum disorder found that all participants successfully acquired abduction-prevention skills, maintaining them at a one-month follow-up.

Critically, research published in Behavior Analysis in Practice (Rodriguez et al., 2020) extended BST to specifically include a safe-word component — teaching children a 4-part response to lures from both strangers and familiar adults. All five child participants met mastery criteria and maintained skills at two-month follow-up. This research directly supports the family code word strategy described below.

Strategy 1: The Family Code Word

What It Is

A family code word is a pre-agreed, secret password known only to household members and a small circle of deeply trusted adults. Any adult who claims to have been sent by mum or dad to collect a child must be able to state the code word. If they cannot — regardless of how convincing their story sounds, regardless of how familiar they seem, regardless of what technology they use to support their claim — the child does not go.

The FBI itself now recommends families establish a code word specifically to protect against virtual kidnapping and AI voice cloning scams, noting that a code word “stops cold” attempts to deceive family members over the phone.

Why It Works

The code word system succeeds because it creates a verification step that no amount of AI voice replication, fabricated photos, or social engineering can bypass. An AI-cloned voice of a parent can sound genuine. A deepfake photo can look real. But neither can produce a word that only the real family knows.

In 2011, this system was put to the test in Ogden, Utah, when a man approached two 8-year-old boys claiming their parents had sent him. When the boys asked for the code word and he could not provide it, they ran. Police later confirmed the man was attempting an abduction.

How to Set It Up

Choose a word that is memorable but not obvious — avoid pet names, family members’ names, or common household words
Share it with your child in a calm, matter-of-fact way — frame it as a special family tool, not a fear-based rule
Include trusted adults (grandparents, regular childcare providers) if needed
Practise using it through low-stakes role play before any emergency arises
Remind children they should never share the code word with peers or anyone outside the designated circle
Refresh the code word periodically, especially if you suspect it may have been overheard or shared

Strategy 2: Teaching Children a Clear Crisis Response

The Limits of Telling Children to “Just Run”

Research shows that verbally instructing children about what to do in danger is far less effective than practised simulation. Children who receive only verbal instruction often freeze, comply, or fail to generalise the skill to real situations. The research on BST consistently finds that rehearsal — actually practising the physical response — is the component that produces durable, reliable behaviour.

What to Teach

When a child senses danger, cannot verify a code word, or feels unsafe, the sequence that research and child safety experts recommend includes:

Do not engage — no negotiating, no listening to explanations, no accepting offers
Create distance immediately — move away from the person with purpose and speed
Make noise loudly and specifically — not just screaming, but stating clearly: “Help! This is not my parent!” or “I don’t know this person!” Specific verbal cues attract more adult attention than generic screaming
Move toward a trusted adult or a crowded place — teach children in advance where safe locations are (shops, school offices, neighbours’ homes, public areas with other adults present)

Cleveland Clinic child health specialists advise that role-play and scenario practice are the most effective way to build these responses into children’s instinct.

Scenario: What This Looks Like in Practice

Eight-year-old Maya is waiting outside her school. A woman approaches and says, “Hi sweetheart, your mum had to work late. She asked me to pick you up today — I’m her friend from work.” Maya pauses. The woman seems kind and knows her mother’s name. But Maya remembers what her family practised. She takes a step back and says, “What’s our family code word?” The woman hesitates. Maya turns, walks quickly toward the school office, and tells the teacher what just happened.

This scenario illustrates exactly what the research predicts: children equipped with both a verification tool (the code word) and a practised physical response are far more likely to act safely than those who have only been told the rules.

Every image posted online becomes a permanent, searchable asset. AI voice cloning requires only seconds of audio — audio that exists in videos, reels, and public stories posted by well-meaning families every day. AI image manipulation requires only a clear facial photo — the kind shared with pride after a school concert or a beach holiday.

Child protection researchers and the Barclays Bank study project that by 2030, so-called “sharenting” — parents sharing images and details of their children online — could account for two-thirds of youth identity fraud cases. This is not an argument against sharing altogether; it is an argument for thoughtful, purposeful sharing with strong privacy settings and a clear-eyed awareness of risk.

Practical Steps

Set all personal social media accounts to private and audit your followers regularly
Avoid posting images that reveal your child’s school uniform, school name, suburb, or daily routine
Avoid sharing videos in which your child’s voice is clearly audible and identifiable, particularly on public-facing platforms
Before posting, ask: If someone with bad intentions found this, what could they learn or do with it?

Strategy 4: Ongoing, Age-Appropriate Conversations

Why One Talk Is Never Enough

Child safety education is not a single event — it is an ongoing, evolving conversation that grows with the child. Research consistently shows that children’s safety understanding deepens with repeated exposure, discussion of scenarios, and age-appropriate updates to the information they receive.

Body Safety Training (BST), an evidence-based programme developed for children aged 3 to 7, emphasises ten short repeated lessons covering multiple safety domains — because repetition across varied contexts is what builds generalisation. The same principle applies to AI-era safety education.

What Age-Appropriate Conversations Look Like

Age Group	Focus Areas	Sample Language
Under 5	Trusted adult circle, body autonomy, always ask before going anywhere	“We have a special family word. If someone needs to pick you up, they’ll know it.”
5–8	Code word, responding to lures, identifying safe adults and places	“If someone says mum sent them, what do you ask first? Let’s practise.”
9–12	AI voice scams, social media risks, response escalation	“Someone on the phone can sound exactly like me now — so our code word matters more than ever.”
Teens	Digital privacy, AI deepfakes, supporting younger siblings	“Even a 3-second voice message can be cloned. Here’s why we keep our privacy settings strict.”

The Bottom Line

The threats children face have become more technically sophisticated, but the protective strategies remain rooted in the same principles: trusted relationships, practised responses, and ongoing dialogue. The addition of the family code word — a simple, low-cost, high-impact tool — directly addresses the newest layer of risk that AI technology introduces.

Preparation is not about raising fearful children. It is about raising children who feel confident, capable, and equipped to navigate the world as it actually is.

Key reports, articles and advisories

Federal Bureau of Investigation (FBI). (2025). Criminals using altered proof-of-life media to extort victims in virtual kidnapping scams (Public Service Announcement). Internet Crime Complaint Center (IC3). Retrieved from https://www.ic3.gov/PSA/2025/PSA251205
Federal Bureau of Investigation (FBI). (2025). FBI sounds alarm over virtual kidnapping scams using social media and AI.
UNICEF. (2026, February 3). ‘Deepfake abuse is abuse’, UNICEF warns. UN News.
United Nations / UN News. (2026, January 25). From deepfakes to grooming: UN warns of escalating AI threats to children.
National Center for Missing and Exploited Children (NCMEC). (2025, August 4). NCMEC releases new data: 2024 in numbers.
ChildLight. (2025). Study finds millions of children face sexual violence – and AI deepfakes surge is driving new threats.
Cleveland Clinic. (2024, January 30). How to teach your kids about “stranger danger” (without scaring them).
Crime Stoppers Australia. (2021, May 30). Tips to keep your child safe.
Nemours KidsHealth. Preventing abductions.
Office of Juvenile Justice and Delinquency Prevention (OJJDP). (2011). Child safety is more than a slogan: “Stranger danger” warnings not effective for keeping kids safer.
ABC News. (2017, March 30). Experts warn against teaching the phrase “stranger danger”.

AI, voice cloning and virtual kidnapping

IC3 / FBI. (2025). Criminals using altered proof-of-life media to extort victims in virtual kidnapping for ransom scams.
axios. (2025, December 5). AI-generated kidnapping scams are coming, FBI warns.
Bitdefender. (2025, December 7). FBI sounds alarm over virtual kidnapping scams, social media extortion and AI.
PK Tech. (2026, March 15). AI involved in kidnapping scam: Here’s what parents need to know.
Grab The Axe. (2026, January 2). The AI-powered terror targeting families and how to verify safety.
Trend Micro. (2023, June 27). How cybercriminals can perform virtual kidnapping scams using AI.
Van Hooijdonk, R. (2024, February 25). The rise of AI voice manipulation in virtual kidnapping.
InvestigateTV+ / Gray Television. (2026, January 27). How AI voice cloning scams target families with fake kidnapping calls.

Generation Next / eSafety Commissioner (Australia). (2015, October). Millions of social media photos found on child exploitation sharing sites.
Digital Watch Observatory. (2025, August 11). Parents should rethink sharing children’s photos online.
Childhood First / tri.x. (2022, May 16). Safe use of the internet, social media and photographs.
ABC News Australia. (2024, August 19). Social media concerns for children in new research.

Behavioural Skills Training (BST), abduction-prevention and safe words

Rodriguez, C. N., & Jackson, M. L. (2020). A safe-word intervention for abduction prevention in children with autism spectrum disorders. Behavior Analysis in Practice, 13(4), 872–882. https://pmc.ncbi.nlm.nih.gov/articles/PMC7666251/
Johnson, B. M., et al. (2006). A preliminary evaluation of two behavioral skills training procedures for teaching abduction-prevention skills to schoolchildren. Journal of Applied Behavior Analysis.
Miltenberger, R. G., et al. (2009). Teaching abduction prevention skills to children: A review and evaluation of training procedures.
Bergstrom, R., Najdowski, A. C., & Tarbox, J. (2014). Teaching children with autism to respond to abduction lures.
How To ABA. (2024, December 14). Episode 133: The BST model – Behavioral Skills Training.
ABA Inside Track. (2018, March 16). Technology and safety skills training.
Organization for Autism Research. (2011, June 27). Teaching children with autism how to respond to the lures of strangers.

Family code words and safety passwords

Rodriguez & Jackson. (2020). A safe-word intervention for abduction prevention in children with autism spectrum disorders. Behavior Analysis in Practice.
Missouri Department of Social Services. Child safety and code words – Safety tip of the month.
Yahoo Parenting. (2017, April 11). Why every family needs a secret code word.
Frankton Police (New Zealand Police Facebook). (2025, January 13). A simple yet effective safety measure: family code words.
Various child safety educators. (2024–2025). Family safety code word videos and resources.

Stranger danger, child autonomy and body safety

Vanderbilt Kennedy Center. (2022, April 6). Teaching stranger safety skills to children and adults with disabilities.

OJJDP. (2011). Child safety is more than a slogan: Stranger danger warnings not effective for keeping kids safer.

NCJRS. (1996). Stranger danger: What do children know? National Criminal Justice Reference Service.

Cleveland Clinic. (2024). How to teach your kids about stranger danger.

Washtenaw Area Council for Children. (2014). Body Safety Training (BST) program.

Wilson, S. R. (2023, July 26). Bodily integrity and autonomy of the youngest children: A rights-based approach.

ABLE. (2021, June 22). Teaching stranger danger.