AI Powers New Wave of Sophisticated Investment Scams
“Main Street fraudsters” are finding ways to broaden the reach of their scams through the use of artificial intelligence, according to state securities regulators.
In a call about the North American Securities Administrators Association’s 2025 Enforcement Report, NASAA Enforcement Section Co-Chair (and Alabama Securities Commission Director) Amanda Senn said financial institutions are reporting AI-powered scams that are so sophisticated that the financial professionals often don’t even realize scammers are impersonating clients.
Additionally, Senn said the proliferation of scam assistants, such as FraudGPT or WormGPT, makes it easier than ever for fraudsters to target investors at scale.
“A fraudster’s no longer required to have a certain level of cyber sophistication; FraudGPT and WormGPT are sold on the black market,” Senn said. “You can get subscriptions, and the application will basically put together your fraudulent scheme or any type of criminal activity.”
So-called “dark” large learning models are mirror-image versions of typical AI systems like ChatGPT reconfigured for criminal purposes, according to The Conversation. Scammers will “jailbreak” regular LLMs by using prompts to circumvent a model’s typical safeguards. The “services” are often advertised on the dark web and on the messaging app Telegram.
According to Senn, the vast amount of information on social media makes it easier for fraudsters to target thousands of victims simultaneously, bringing “volume, scale and sophistication” to what previously may have been small-scale operations.
While fraudsters often target financial institutions by claiming to be clients (or someone to be legitimately added to a client’s brokerage or investment account), Senn said investors also get calls from their agent, representative or bank about transactions (which turn out to be AI-generated fakes), which result in “millions of dollars of losses.”
According to the NASAA report (which analyzed 2024’s findings), digital assets and cryptocurrencies ranked as investors’ leading threats for the third consecutive year, along with “pig butchering” scams in which fraudsters build trust with victims before trapping them in scam investment opportunities.
Last year, tips and complaints to state securities regulators increased to 8,309, in addition to referrals from agencies like FINRA and the Securities and Exchange Commission (advisors and firms managing under $100 million in assets are state-regulated).
Regulators conducted 8,833 “active” investigations, with $259 million in fines and restitution. These included 463 cases involving digital assets and 229 instances of pig butchering, among other types of crimes. Senn stressed that pig butchering schemes didn’t always involve an “element of romance,” claiming regulators saw schemes centered on alleged job opportunities or products that required a down payment.
Additionally, the blockchain made digital asset schemes particularly difficult, as fraudsters can remain anonymous more easily than in traditional banking networks.
“It’s very tedious and time-consuming to investigate any type of matter that involves a digital asset,” Senn said, noting that in Alabama, it took a program designed to trace these kinds of frauds “upwards of 16 hours” to track one pig butchering scheme.
These schemes have continued amid federal regulatory workforce reductions at agencies like the SEC since the start of the Trump administration and an ongoing government shutdown, which has resulted in all but approximately 9% of SEC workers being furloughed.
According to Senn, it was “too soon to tell” what impact the federal workforce reductions at (as well as the shutdown) would have on state securities regulators, while also noting that the amount of fraud leaves a gargantuan workload for them regardless of federal developments.
“But this time next year, we’ll see what the regulatory landscape looks like and I’ll know from talking with my colleagues whether they’re dying over the number of cases pushed down,” Senn said.
Post Comment