You get a panicked call from your grandchild. They've been in an accident, or arrested, and need money — fast. The voice sounds unmistakably like them. Except it isn't. With as little as three seconds of audio pulled from a public social media video or voicemail greeting, AI tools can now clone anyone's voice convincingly (FTC Consumer Alert, March 2023).

In February 2025, federal prosecutors indicted 25 suspected members of a Canadian fraud ring accused of stealing more than $21 million from grandparents across 46 U.S. states using AI voice cloning.

Extreme urgency · Demand for secrecy ("don't tell Mom") · Request for gift cards, wire transfers, or cryptocurrency

Hang up immediately. Call your loved one back at their real number. Better yet — agree on a family safe word today. A random phrase only your family knows is the single most effective defense recommended by the FBI, FTC, and AARP.

What used to take Hollywood studios now takes a laptop. In early 2024, an employee at the global engineering firm Arup transferred $25.6 million after attending a video call with what appeared to be the company's CFO and colleagues — every one of them an AI deepfake (CNN Business, May 2024). The same technology is now being aimed at individuals through fake "emergency" video calls.

Slight delays in lip sync · Faces look smooth or "too perfect" · Unusual requests during a video call · Can't respond naturally to spontaneous questions

Ask the person to turn their head sideways or pick up a nearby object. Deepfake software still struggles with profile views and unexpected movement. Always verify large financial requests through a second, known channel — a real phone call to a number you already have.

Tech support scams are the #1 fraud reported to the FBI by adults 60 and over, costing victims $982 million in 2024 alone (FBI IC3 2024 Report). AI chatbots now power realistic "Microsoft," "Apple," or "Geek Squad" pop-ups and live chats that walk you through giving a scammer remote access to your computer. The FTC reports that Best Buy/Geek Squad, Amazon, and PayPal are the brands most often impersonated.

Unexpected pop-ups claiming your computer is infected · Calls from "Microsoft" or "Apple" · Requests to install remote-access software · Instructions to pay in gift cards or Bitcoin

Microsoft, Apple, and Amazon will never call you unsolicited. Close the browser, restart your computer, and call a trusted local tech professional — not a number from the pop-up.

Romance scams cost older Americans more than $390 million in 2024 (FTC Protecting Older Consumers Report, 2024–2025). Today's romance scammers use AI-generated profile photos that can pass reverse-image searches, and AI chatbots to maintain conversations with dozens of victims simultaneously. AARP research found that scams identified as AI-enabled increased twenty-fold between 2023 and 2025.

Falls in love very quickly · Always has an excuse not to video chat · Eventually needs money for a medical emergency, customs fee, or "investment opportunity" · Claims to be overseas (military, oil rig, etc.)

Never send money to someone you haven't met in person. Run their profile photos through Google Images reverse search. Talk to a trusted family member or friend before acting on any online relationship.

Scammers use AI to spoof caller ID, generate fake government letterhead, and produce voicemails that sound like real IRS or Social Security agents. Older adults who lost more than $100,000 to these impersonation scams accounted for combined losses of $445 million in 2024 — an eight-fold increase since 2020 (FTC, August 2025). The FTC's Government and Business Impersonation Rule took effect in April 2024 specifically because these scams cost Americans $2.95 billion in 2024.

Anyone telling you to "move your money to protect it" · Requests to buy gold bars · Instructions to deposit cash at a Bitcoin ATM · Demands to hand cash to a courier

The IRS, FBI, and Social Security Administration will never do any of the above. Hang up. Call the agency directly using a number from their official .gov website.

In March 2023, Samsung allowed its semiconductor engineers to use ChatGPT to help debug code. Within 20 days, the company identified three separate incidents where engineers had pasted in confidential source code, internal meeting notes, and equipment test data — all of which OpenAI's servers could potentially retain. Samsung banned generative AI company-wide shortly after (Bloomberg, May 2, 2023).

The lesson: Without a clear "never paste this" rule, even careful, well-meaning employees can expose your most sensitive information.

In January 2024, a finance employee at global engineering firm Arup joined a video call he believed was with his CFO and several colleagues. Every face on the call was an AI deepfake. He made 15 transfers totaling $25.6 million before the fraud was discovered (CNN Business, May 2024).

The lesson: A written policy requiring a second verification channel for any wire transfer — a real phone call to a number you already have — would likely have stopped this.

The FTC's Operation AI Comply initiative has already targeted businesses using AI chatbots to provide fake legal advice and post phony reviews — with penalties of up to $53,088 per violation under the Government and Business Impersonation Rule. If your AI-generated marketing content misleads consumers, your business is the one that gets fined, not OpenAI.

The lesson: "We didn't know what our AI was doing" is not a legal defense.

Don't try to "transform your business." Pick the most boring, repetitive thing on your desk right now: drafting follow-up emails, summarizing long documents, writing product descriptions, or building a first draft of a proposal. Ask the AI to do it. Edit the result. Track how much time you saved.

A Goldman Sachs research analysis found that knowledge workers using AI on specific, well-matched tasks recorded median productivity gains of around 30% for those use cases. A Harvard/MIT study of GitHub Copilot found developers completed coding tasks 55.8% faster with AI assistance. The wins are real — they just show up in specific use cases, not across everything you do.

The single biggest predictor of AI success isn't the tool — it's whether someone actually learned to use it. McKinsey's 2025 State of AI report found that organizations seeing real bottom-line impact from AI are the ones where leaders redesign workflows and provide role-based training.

Free training options: Microsoft Learn, Google AI Essentials (~$49 on Coursera), OpenAI's help center, and your local SCORE chapter all offer beginner courses. Two hours of focused practice will take you further than two months of reading articles about AI.

Before you let multiple employees use AI on customer data, spend 30 minutes writing down two things: what they can use AI for, and what they should never paste into it (client names, financial data, medical records, anything under NDA). That's the foundation of an AI policy — and it prevents the Samsung-style leaks and Arup-style wire fraud described in our governance article.

See our full guide: Does Your Small Business Need an AI Policy?