AI carries security risks in banking, including being used by scammers to target financial information. Photo via Getty Images

With artificial intelligence technologies easily accessible and growing in popularity, consumers and business owners alike should be aware of both the benefits and risks when it comes to the utilization of generative AI tools in banking and finance. While data-driven AI creates the opportunity to further drive innovation in banking, the data-reliant nature of the industry makes it a natural target for scammers looking to intercept personal and business finances and sensitive customer information.

As banks and other financial service providers are using AI as a tool to scan for anomalies or errors that are known fraud techniques, criminals are using AI to improve their chances of perpetrating fraud. For this reason, consumers and businesses should guard their data with the same diligence used to guard cash and other valuable physical property.

Privacy and accuracy

For entrepreneurs and businesses of all sizes, it is important to keep in mind the practical applications of AI beyond the trending headlines, whether implementing the technology into everyday internal business practices, or into client-facing solutions.

When feeding information into AI, it is best to maintain a defensive position and be proactive about not disclosing sensitive or private information. Also, rely on sound judgment when deciding when and how to use AI technologies. From a business standpoint, privacy should be embedded into a financial system’s design and leaders should be transparent about the technologies used within a given system.

Technologies like ChatGPT are large language models operating on massive datasets, including documents and web pages across the internet. This poses a risk because some sources of this data lack accuracy. When seeking financial advice via AI technologies, it is best to conduct research by curating and limiting the dataset then talking through your unique financial position in person with your trusted banker and IT staff or consultants.

Phishing and business email compromise via AI

Historically, phishing and business email compromise, or BEC, attempts have been more easily recognizable and often flushed out due to grammatical errors and unnecessary punctuation. With technologies like ChatGPT, scammers are now better equipped to draft well written content that can fool a person into thinking a communication is legitimate. Phishing can lead to people clicking links or attachments that harbor malware or other viruses that can lead to account takeover. With BEC, a person might be fooled into thinking an email is from a legitimate person. Scams like these could potentially lead to the disclosing of sensitive information or accepting transaction instructions or changes, ultimately resulting in money being sent to a fraudster.

AI voice generators

AI voice generators can be used to mimic voices of anyone including bankers, C-suite leaders and customers. If a person is fooled into believing they have received a voicemail or are talking to a person they know, they may accept instructions from a fraudster like providing transaction approvals and sensitive or private information, resulting in fraud.

AI can also create fake identities, including AI-developed photos of individuals, and other false information. These fake identities could be used to create accounts for fraudulent purposes.

AI is here to stay

AI is forecasted to have a lasting impact on the banking industry. Whether on the business or consumer side of the spectrum, it will be important to embrace the innovation and enhancements generative AI will continue to produce, while maintaining a cautionary stance around protecting client and business information and finances. Fraud prevention practices will need to continue evolving alongside the fast-paced growth of generative AI in banking.

------

Ken Smiley is treasury management division manager of Amegy Bank and a fraud protection expert.

Ad Placement 300x100
Ad Placement 300x600

CultureMap Emails are Awesome

Houston hardtech accelerator names 8 founders to 2026 cohort

hardtech fellows

Hardtech-focused organization Activate has named 50 new members to its 2026 cohort of scientists, which includes eight startups joining Activate Houston.

Activate aims to support scientists at "the outset of their entrepreneurial journey." It partners with U.S.-based funders and research institutions to support its fellows in developing high-impact technology. Its fellows receive a living stipend, research and development funding, connections from Activate's robust network of mentors and access to a curriculum specific to the program for two years.

This year's fellows represent 41 companies from 22 U.S. cities and 11 states.

“This cohort clearly demonstrates that the next industry-defining companies won't choose between modern technology and deep science; they'll be built by combining both,” Cyrus Wadia, CEO of Activate, said in the announcement. “These are the scientists and engineers turning our most urgent global challenges into the companies that will deliver a more sustainable future.”

The Houston fellows are working across the energy, space, AI infrastructure and agriculture sectors. They include:

  • Sophie Clare Broun, founder of Anning Corporation, which is producing clean hydrogen by stimulating naturally occurring geologic deposits
  • Kathy Andersen, founder of Brint Tech, which builds optical sensing systems that quantify hydrogen for infrastructure operators
  • Dorsa Talebi, founder of Kinetiq Drive, which builds rare-earth-free, contact-free electric motors with wireless rotors for small appliances and heavy industry alike
  • Neethu Pottackal, founder of Nivera, which is developing natural, edible coatings made from agricultural byproducts to reduce food waste and extend the shelf life of fresh food
  • Jonathan Huffman, founder of Orbital Arc, which is shrinking spacecraft propulsion to a microchip powerful enough for deep space
  • Tim Lee, founder of Renesin, which develops advanced materials for faster, more efficient AI hardware
  • Joshua Livingston, founder of Selerra Separations, which is developing high-performance membranes that cut the cost and energy consumption of water treatment
  • Wenli Jiang, founder of SwieNitro Recovery, which is developing technology that converts nitrogen-rich waste streams into valuable fertilizer

"Home to the largest concentration of engineers in the United States and a dense ecosystem of Fortune 100 and Fortune 500 companies, Houston is uniquely positioned for scientists tackling large-scale industrial challenges. Activate Houston fellows are connected to the city's deep networks in energy, chemicals, and materials," Activate said in the announcement.

Activate named its inaugural Houston cohort in 2024. It has other hubs in Boston, New York, and Berkley, California—where Activate is headquartered. The organization also offers a virtual and remote cohort, known as Activate Anywhere. Nationally, it has supported 346 fellows and 276 companies since 2015.

Activate Houston is led by managing director Jeremy Pitts, who co-founded Greentown Labs in Boston. It is based out of the Ion. The latest cohort is Activate Houston's third. Read more about the last year's cohort here.

Rice, Houston Methodist award $90K for cancer research projects

seed grants

Rice University’s Synthesis X Center, in partnership with the Houston Methodist Neal Cancer Center, announced earlier this month the organizations would award $90,000 in seed grant funding to two projects that could help fight cancer. One would ease the pain of chemotherapy, while the other could change the way the progression of leukemia is tracked.

“We’re excited to be collaborating with the Neal Cancer Center to support collaborative, novel and interdisciplinary proposals to improve cancer care and outcomes,” Han Xiao, Rice professor of chemistry and the director of the SynthX Center, said in a news release. “Together, we can achieve translational excellence.”

The projects

Chemotherapy is one of the most effective cancer treatments, but it can be hard on the body. The human body doesn’t like being injected with radioactive material, especially the skin around the injection point, which can become extremely irritated. For patients in long-term treatment, the skin irritation can be more than just a bother; it can lead to infections that are dangerous to a compromised immune system.

Angel A Marti, a professor of chemistry at Rice University, and Biana Godin, an associate professor of nanoscience at Houston Methodist Research Institute, are experimenting with metal nanoclusters as a way to block radiation at the injection site. The nanocluster could be applied in a cream or a gel on the skin, serving as a type of shield against the harsh radioactive material.

Meanwhile, Yuan Ma, an assistant professor of chemistry at Rice, and Shu-Hsia Chen, a professor of immunology at Houston Methodist Research Institute, are working with m6A. Discovered in the 1970s, m6A is the most prevalent chemical modifier found in mRNA in mammals. It is prevalent in many cancers, including leukemia.

Ma and Chen are working on measuring the amount of m6A in leukemia to see if it can determine the most effective cancer treatments. The team is also experimenting with ways to shut off m6A to see if it makes current leukemia treatments more effective.

Progress from SynthX

SynthX was first launched in April 2024 to turn research from Rice and Houston Methodist into real-world cancer treatments. Within a year, the center had secured $1.5 million in grant money to work on crossing the blood-brain barrier in brain cancer treatments. These latest awards show that SynthX Center continues to bridge the worlds of research and clinical practice.

“This collaboration reflects a shared commitment to team science in cancer research,” Daniela Matei, the director of the Houston Methodist Neal Cancer Center, added in the release. “These are the types of translational and pioneering projects that lead to transformation in patient care.”'

The SynthX Center awarded $90,000 in seed grants to three teams in 2025 and $80,000 to three teams in 2024.

Tesla self-driving mode wasn't to blame in Houston-area crash, report suggests

Tesla news

Federal safety investigators looking into a runaway Tesla that killed a grandmother in her home say the driver had pressed the accelerator to full speed, suggesting the vehicle's self-driving software was not to blame.

The driver had told police that he had the self-driving software turned on, but a report from the National Transportation Safety Board concluded that he had actually overridden that feature when he pushed hard on the pedal. Moments later the Tesla Model 3 raced down a residential street in Katy, Texas, at highway speeds, slammed into a brick home and killed a 76-year-old woman standing in the front room.

The crash last month drew national attention because Tesla CEO Elon Musk is seeking to reassure the public its self-driving feature is safe as he prepares to turn hundreds of thousands of Teslas already on the road into fully automatic vehicles and begin selling two-seated Cybercabs missing steering wheels and pedals.

The crash came two months after officials at a separate federal agency, the National Highway Traffic Safety Administration, announced it was elevating a 2024 investigation of the self-driving feature to new “engineering analysis” level, raising the possibility of a recall of 3.2 million Tesla vehicles.

That NHTSA probe was triggered by crashes where the self-driving feature failed to alert drivers to take control in fog and other poor visibility conditions.

The agency opened an investigation last year into 58 incidents in which Teslas reportedly violated traffic safety laws while using self-driving technology, leading to more than a dozen crashes and fires and nearly two dozen injuries.

Separate from the National Transportation Safety Board, NHTSA is also looking into the Tesla house crash in Texas, one of 46 “special crash” investigations of Tesla's self-driving or driver-assistance technology in the past decade, according to the agency’s records. In more than a dozen of those crashes, at least one person — a driver, passenger or pedestrian — was killed.

Tesla had originally called its driver assistance software Full Self-Driving, or FSD, but auto experts and regulators complained it was misleading because drivers must always keep their eyes on the road and be ready to take over at any time.

The company has since changed the name to Full Self-Driving (Supervised).

Video of the Katy, Texas, accident shows the Tesla traveling at more than 70 mph (112.65 kilometers per hour), jumping a curb then tearing across a lawn before crushing through a brick wall of a home. A woman standing feet away, Martha Avila, was found amid piles of crumbling plaster, split beams and bits of furniture and rushed to a hospital but died.

Sales of Tesla cars still haven't recovered fully from boycotts last year over Musk's political stands, but the stock is rising anyway as he has successfully shifted attention away from the sales figures. He says they matter less now that the company is on the cusp of major technological advances, such as turning Teslas into hands-free vehicles and having its Optimus robots take over for humans for tasks at home and work.

Tesla stock has risen 22% in the past year and is currently trading at 170 times expected annual earnings compared to 20 for the S&P 500.

For its second-quarter financial results, financial analysts surveyed by FactSet expect earnings per share will barely budge — 32 cents versus 33 cents a year earlier — continuing a sixth quarter streak of flat or falling profits.