Fraud may be emerging as one of the most effective uses of artificial intelligence, according to a new study.
In a simulated "pig butchering" scam, an AI chatbot outperformed human scammers, and participants even recalled they trusted the bot more.
Here's what to know
Researchers from four universities posted arXiv findings from a weeklong experiment involving 22 unsuspecting participants that compared an AI chatbot with human scammers, VICE reported.
In the test, VICE said, each participant texted with two strangers: one was a human scammer, and the other was a bot instructed to seem human and not reveal it was AI.
Rather than a simple cash grab, the experiment was modeled on a "pig butchering" scam, according to VICE — a scheme in which fraudsters may spend weeks or months posing as friends or romantic partners before eventually directing victims toward questionable investments.
VICE said successful compliance was much higher with the chatbot: 46 percent of participants installed an app at its request, compared with 18 percent who did so for the human scammer.
TCD Picks » Quince Spotlight
💡These best-sellers from Quince deliver affordable, sustainable luxury for all
Roughly four-fifths of participants' messages went to the chatbot, according to VICE, and they also said they trusted it more than the human scammer.
VICE reported that the bot's advantage appeared to stem from its ability to keep conversations moving, quickly recall personal details, and remain consistently engaged with its target.
More background
"Pig butchering" scams have already become a major threat because they rely on an unfortunately effective method of emotional manipulation — rather than obvious technical tricks.
The scam usually does not start with an immediate demand for money. Instead, the scammer first builds a sense of personal connection, making the eventual pitch feel more believable and less suspicious.
The study suggests that scammers' ability to build a sense of trust with victims could become scalable through automation, potentially leading to more severe consequences. If AI can take over the slow work of maintaining relationships, scammers may be able to reach more targets for cheaper and with fewer clear warning signs.
That could make these schemes even more widespread and harder to trace.
Some scam operations in Southeast Asia depend on trafficked workers who are forced to carry out fraud in scam compounds, VICE noted.
Broader use of AI could reduce that exploitation in some cases while also making scams more efficient and more common.
AI can help optimize electric grids, improve forecasting, and support cleaner energy systems by matching supply and demand more efficiently.
At the same time, training and running AI models can require enormous amounts of electricity and water for data centers, potentially straining local grids, raising costs, and creating new security and social risks if the technology is misused.
What can be done?
Slowing down before making any risky decisions can be a strong defense against these types of scams.
Online-only relationships that quickly become emotionally intense or shift toward investing, crypto, private apps, or urgent financial decisions warrant skepticism.
Downloading unfamiliar apps, moving money, or sharing financial information based on the recommendation of someone you have never met in person can also be a major red flag.
Independent verification through other channels can help, as can running any investment pitch by a trusted friend, family member, or licensed financial professional.
At a broader level, the study points to the need for stronger safeguards around AI impersonation, fraud detection, and platform monitoring.
If bots can convincingly mimic human warmth and patience, companies and regulators will need tools to identify and disrupt such interactions earlier.
Get TCD's free newsletters for easy tips, smart advice, and a chance to earn $5,000 toward home upgrades. To see more stories like this one, change your Google preferences here.
