A grandmother in Texas picks up the phone and hears her grandson’s voice, panicked, saying he has been in a car accident and needs bail money wired immediately. The voice is his, down to the slight drawl and the way he rushes through sentences. Except it is not him. It is a synthetic replica generated by artificial intelligence from a short clip pulled off social media. By the time she realizes something is wrong, thousands of dollars are gone.
Variations of that scenario are playing out across the country at an alarming pace. Fidelity Investments, which manages more than $5.8 trillion in assets, has flagged what it describes as a $15 billion fraud wave targeting Americans’ savings. According to Fidelity’s customer alerts and fraud education materials circulated in early 2025, scammers are now blasting an estimated 100,000 deceptive text messages per day and using AI voice-cloning tools that can produce a convincing replica of someone’s voice from as little as three seconds of recorded audio.
Fidelity has not published a detailed methodology behind the $15 billion figure. However, the number falls between the two most authoritative public benchmarks for 2024 fraud losses, both of which are widely considered undercounts because they rely on voluntary victim reports. That context is important for understanding why a large asset manager tracking millions of daily transactions might arrive at a higher estimate than what federal complaint data alone can capture.
Federal data confirms the surge
The Federal Trade Commission reported in March 2025 that consumers lost more than $12.5 billion to fraud in 2024, a 25% jump over the prior year based on roughly 2.6 million reports. Investment scams led all categories at $5.7 billion. Imposter scams, the kind where criminals pose as bank representatives, government officials, or family members, accounted for $2.95 billion.
The FBI’s Internet Crime Complaint Center went further. Its 2024 annual report tallied $16.6 billion in total internet crime losses and explicitly named cryptocurrency fraud and AI-powered impersonation as accelerating threats. Americans over 60 bore the heaviest burden, losing more than $4.8 billion.
Both agencies are clear that their numbers capture only what victims choose to report. The FTC has long noted that most fraud goes unreported, which means $12.5 billion is almost certainly a floor. That gap between reported and actual losses is a key reason industry estimates like Fidelity’s run higher, and why fraud researchers treat even the FBI’s $16.6 billion as conservative.
How AI voice cloning works in practice
The technology behind voice cloning has moved from academic research to criminal toolkits faster than most consumers realize. In 2023, Microsoft demonstrated a text-to-speech model called VALL-E that could generate a realistic vocal replica from a three-second sample. OpenAI has shown similar capabilities. Those tools, or open-source versions of them, are now accessible to anyone with a laptop and basic technical skills.
Scammers harvest voice samples from sources most people never think to protect: voicemail greetings, TikTok videos, Instagram stories, YouTube clips, even brief phone calls where the target says a few words before hanging up. Once the clone is built, the attacker calls a relative, employer, or financial institution using the synthetic voice, often spoofing the caller ID to match the real person’s number.
The scripts are engineered around panic. A “grandchild” begs for emergency bail money. A “spouse” says the family’s bank account has been compromised and funds need to be moved right now. A “boss” instructs an employee to wire money to a vendor. The emotional pressure, paired with a voice that sounds authentic, overwhelms the skepticism that might otherwise stop someone from complying.
The FBI has confirmed that AI-enabled impersonation is now part of the standard fraud playbook. Federal agencies have not yet published a standalone breakdown of losses tied specifically to voice-cloning attacks versus other AI-assisted techniques like deepfake video or AI-generated phishing emails. That data gap makes it hard to isolate exactly how fast voice cloning is growing, but cybersecurity researchers and law enforcement officials describe the trajectory as steep.
The text message flood
Voice cloning gets the attention, but high-volume text message scams remain the workhorse of the fraud economy. The 100,000 deceptive texts per day cited in Fidelity’s warning is consistent with broader industry data. The anti-spam firm RoboKiller estimated that Americans received more than 225 billion spam texts in 2024, a figure that includes marketing messages, phishing attempts, and outright fraud lures. Dividing even a small fraction of that total by 365 days yields daily volumes well above 100,000 for messages classified as fraudulent, which helps contextualize Fidelity’s figure even though the firm has not disclosed its own counting method.
The most common text-based scams impersonate package delivery services, banks issuing fraud alerts, and government agencies like the IRS or Social Security Administration. Each message is designed to get the recipient to click a link, call a phone number, or reply with personal information. The texts cost almost nothing to send in bulk, are difficult to trace, and generate significant criminal revenue even with a tiny response rate.
Wireless carriers have deployed filtering tools that block billions of suspected scam messages each year, but fraudsters constantly rotate phone numbers, domains, and message templates to stay ahead. The FCC has pushed carriers to implement STIR/SHAKEN caller-authentication protocols, which help verify the origin of voice calls but do less to stop text-based schemes.
Why retirement and savings accounts are prime targets
Fidelity’s warning zeroes in on the threat to savings for a simple reason: retirement accounts hold concentrated wealth that many owners rarely monitor. A 401(k) or IRA with a six-figure balance may go weeks without the owner logging in, giving a scammer who gains access a window to initiate withdrawals or redirect distributions before anyone notices something is wrong.
Imposter scams targeting older adults are especially effective because many retirees are less familiar with what AI can now do and more trusting of phone calls that appear to come from known contacts. The FBI’s finding that Americans over 60 lost $4.8 billion in 2024 underscores how heavily this age group is targeted.
Major financial institutions have responded by adding verification layers for large transactions, expanding fraud-monitoring algorithms, and publishing customer education materials. Fidelity, for example, encourages customers to set up a verbal password on their accounts and to enable two-factor authentication. Vanguard and Charles Schwab have rolled out similar measures. But the speed at which AI-driven attacks evolve means institutional defenses are perpetually playing catch-up.
How to protect yourself and your family
Security experts and federal agencies offer consistent, practical guidance that does not require technical expertise:
- Enable multi-factor authentication on every financial account, email address, and phone carrier account. A stolen password alone should never be enough to get in.
- Use a password manager to generate and store unique passwords. Reusing passwords across sites remains one of the easiest vulnerabilities for criminals to exploit.
- Treat urgency as a red flag. Legitimate banks and government agencies will not demand immediate action over the phone or by text. If someone pressures you to act now, hang up and call back using a number from the institution’s official website or the back of your card.
- Establish a family code word. Pick a word or phrase that only your household knows. If someone calls claiming to be a relative in distress, ask for the code word before sending money or sharing any information.
- Verify through a second channel. If you receive a suspicious call from someone who sounds like your spouse, child, or boss, hang up and reach that person directly through a different method: a text, a video call, or an in-person check.
- Limit public audio exposure. Review the privacy settings on social media accounts where your voice appears. Scammers mine public videos and voicemails for cloning material.
- Set up account-specific verbal passwords. Many brokerages and banks now allow customers to require a verbal password before any phone-based transaction. If your institution offers this, use it.
Reporting fraud also strengthens the data that regulators and law enforcement rely on to track emerging schemes. Victims and witnesses can file reports through the FTC’s online portal, the FBI’s Internet Crime Complaint Center, or local law enforcement.
Why the old defenses no longer hold
As of mid-2026, the gap between what federal agencies can measure and what the financial industry is seeing on the ground remains wide. The FTC’s $12.5 billion and the FBI’s $16.6 billion are the most reliable public benchmarks for 2024 losses, but both agencies caution that actual harm runs significantly higher. Fidelity’s $15 billion estimate fits within that range and reflects the kind of real-time pattern recognition that large financial firms, sitting on top of millions of daily transactions, are uniquely positioned to perform.
The direction is not in question. Fraud losses have climbed every year for the past half-decade. AI tools are making scams cheaper to run and harder to detect. The volume of deceptive messages hitting Americans’ phones shows no sign of slowing. Defenses that worked five years ago, like recognizing a robotic-sounding voice or spotting a typo-riddled email, are no longer enough against adversaries who can clone a loved one’s voice from a three-second clip and fire off a hundred thousand texts before breakfast.
The most effective countermeasure is still the simplest: pause before you act. Scammers depend on speed and emotion. Slowing down, verifying independently, and reporting what you encounter are the tools that break the cycle, no matter how sophisticated the technology on the other end of the line becomes.
Warren Cohen is a finance writer based in Phoenix, Arizona, covering personal finance topics including credit, banking, and beginner investing. He earned his degree in business administration from Arizona State University and began his career working in consumer finance, where he gained direct experience with lending and credit systems. He now writes for personal finance websites and fintech platforms, focusing on clear, practical content that helps readers make informed financial decisions.
