Reports: Is There an Over-Sharing with LLMs (Like ChatGPT)?, Elder Fraud Increases, and Crypto-Obsessed States

By Roy Urrico

A mixed bag of reports on the fringes of the financial services industry focuses on elder fraud, the most crypto-obsessed states, and whether there is an over-reliance on artificial intelligence (AI) large language models (LLMs)

Sharing Personal Data to Get Help from ChatGPT and other LLMs

Fifty-two percent of U.S. adults use LLMs, like ChatGPT, with a third believing they are now “too dependent” on them for answers, according to a recent survey from Elon (N.C.) University. In the survey, conducted among 500 LLM users between Jan. 21-23, 2025, 38% of American employees also admit to sharing sensitive work information with AI tools without their employer’s permission.

“By any measure, the adoption and use of LLMs is astounding. I am especially struck by the ways these tools are being woven into people’s social lives,” said Lee Rainie, director of the Imagining the Digital Future Center at Elon University. “Our survey shows that the share of those who use the models for personal purposes significantly outnumber those who use them for work-related activities.”

Some of other key findings about LLM usage:

• 51% said the main purpose for their use is personal, informal learning, compared with 24% who said the main purpose is work activities.

• LLM usage breakdown: 72%, ChatGPT; 50%, Google’s Gemini; 39%, Microsoft’s Copilot; 20%, Meta’s Llama; 12%, xAI’s Grok; 9%, Anthropic’s Claude; and 9% other LLMs.

• 54% said use improved their productivity a lot or somewhat.

• 50% said use improved their ability to learn new skills and concepts a lot or somewhat.

Dallas-based Indusface, an application security firm, investigated what personal data Americans might be oversharing with LLMs based on the Elon University survey’s findings.

• Financial information. “LLMs are lacking in numerical literacy as they are primarily a word-processing tool, so inputting financial figures into ChatGPT is likely to output mistakes and potentially harmful business strategies,” suggested Indusface. “It is best practice to use LLMs as an aid in your understanding of finance, rather than a tool to calculate solutions or make important financial decisions.”

• Work files, such as reports and presentations. One of the most common categories of information shared with AI is work-related files and documents. Over 80% of professionals in Fortune 500 enterprises use AI tools, such as ChatGPT, to assist with tasks such as refining emails, reports, and presentations. However, 11% of the data that employees paste into ChatGPT is strictly confidential, such as internal business strategies.

• Passwords and access credentials. LLMs are not designed with confidentiality in mind, but rather the purpose to learn from what users’ input, the questions they ask, and the information they provide.

• Personal details, such name and address. Sharing personal details such as name, address, and recognizable photos makes individuals vulnerable to fraud. “It is critical to avoid feeding LLMs information that might allow fraudsters to either 1) impersonate you, or 2) create deepfakes, which depict people saying or doing something they never said or did,” according to the Government Accountability Office.

• Company codebases and intellectual property (IP). Developers and employees increasingly turn to AI for coding assistance. However, sharing company codebases are a major security risk, as it is a business’s core intellectual property.

Data Exposure Fueled $4.9 billion in Elder Financial Fraud Losses

With financial losses increasing by 43% year over year (YoY) to almost $4.9 billion in 2024, Los Angeles-based Incogni, a data privacy company, studied Internet-based crimes targeting Americans aged 60 and older. The Incogni research analyzed the FBI Internet Crime Complaint Center (IC3)’s 2024 Internet Crime Report, released in April 2025 that examined year-over-year trends and the role of online personal data availability. Of the 113,906 crimes involving elders reported in 2024, 72% were driven by the availability of victims’ personal data online.

Darius Belejevas, head of Incogni.

Online elder fraud increased significantly last year, with the FBI IC3 reporting record-high numbers of complaints and financial losses. Data brokers and ‘people search’ sites contributed significantly to the problem by making personal data (including street and email address, phone number, family members’ names, job history and more) available to fraudsters at a low cost.

Other key Insights of the Incogni analysis:

• Phishing and spoofing were the most prevalent types of cybercrime reported by elders in 2024, reported a total of 23.3 thousand times—a 700% increase YoY.

• Investment scams had the highest loss-to-report ratio and were associated with losses amounting to $1.83 billion, or $194,100 per complaint.

• In 2024, Texas senior citizens suffered the greatest average losses per complaint ($51.7 thousand), followed closely by those residing in Georgia and California, where reported losses per complaint averaged over $48.2 thousand and $46 thousand, respectively

“The fact that 72% of elder‑targeted cybercrimes are enabled by personal data exposure should be a national wake‑up call,” said Darius Belejevas, head of Incogni. “The unchecked circulation of sensitive personal information is fueling a crisis that costs older Americans billions every year. Vulnerable individuals can take preventive measures, such as having their personal information removed from circulation.”

America's Crypto Capitals of 2025

The crypto site Zellix analyzed the states with the highest search volumes for terms related to cryptocurrency, such as “Which crypto to buy today” and “What crypto will explode in 2025.” The methodology included a seed list of initial search terms based on cryptocurrency terms the monthly average search volume for each term in each state as taken from Google Keyword Planner. The total number of monthly average searches in each location helped establish the monthly average searches per 100,000 people.

Findings concluded New York is the state most interested in cryptocurrency, topping the ranking with an average of 298.58 monthly searches for crypto-related terms. Nevada is in second place with an average of 257.63 monthly searches, while California is in third place with an average search volume of 252.43 per 100,000 people.

Experts at Zellix, an online crypto retail platform, decoded some of the most searched-for crypto-related terminology:

Crypto. “What is cryptocurrency?” is the most popular search term when it comes to crypto lingo. According to Google Trends data, worldwide searches for ‘What is crypto?’ have skyrocketed by 288% in recent months.

Blockchain. “What is blockchain?” is another frequently searched-for crypto term, with online searches for this term increasing by 423% over the past 12 months.

Mining. In crypto terms, mining refers to the process by which individuals race to record a transaction onto the blockchain using powerful computers that are programmed to validate transactions.

Crypto wallet. Refers to where an investor holds their digital assets. There are two types of crypto wallets: hot wallets (connected to the internet), and cold wallets (physical items like hard drives and USBs).

Coinbase. According to Google Trends, worldwide searches for ‘What is Coinbase’ have risen by 243% over the past 12 months. Coinbase is one of the leading platforms in the industry for buying, selling, and trading crypto.

Bitcoin. Over the past few weeks, Bitcoin searches have risen by 150%.

Trajan King, the CFO at Zellix, commented: “It is fascinating to see how interest in cryptocurrency differs on a regional level. This can be linked to various factors, including whether the state has a strong culture of financial innovation or a growing technology sector.”