Your financial life becomes training data for the next version of ChatGPT
En un momento que podría redefinir la relación entre la inteligencia artificial y la vida financiera cotidiana, OpenAI ha permitido a los suscriptores de ChatGPT Pro en Estados Unidos vincular sus cuentas bancarias directamente al chatbot, conectándolos con más de 12.000 instituciones financieras. La promesa es seductora: un asistente que comprende el panorama completo de tus finanzas personales y puede orientarte con precisión. Pero esta expansión del poder de la IA hacia uno de los dominios más íntimos de la existencia humana —el dinero— plantea preguntas que van más allá de la tecnología: ¿hasta qué punto estamos dispuestos a confiar nuestros secretos financieros a una empresa cuya transparencia ha sido cuestionada repetidamente?
- OpenAI abrió una puerta sin precedentes: ChatGPT ahora puede ver saldos, transacciones, inversiones y deudas de millones de usuarios, transformando al chatbot en algo parecido a un asesor financiero personal.
- La configuración predeterminada de la plataforma permite que los datos financieros compartidos sean utilizados para entrenar futuros modelos de IA, una cesión silenciosa que muchos usuarios no advertirán a menos que revisen activamente sus ajustes de privacidad.
- Expertos en seguridad advierten que concentrar información financiera tan sensible en un solo sistema crea un blanco de alto valor para ciberataques, fraudes de phishing y esquemas diseñados a medida según los hábitos de gasto de cada usuario.
- La credibilidad de OpenAI está en entredicho: cambios reiterados en sus políticas de datos, presiones para aumentar la rentabilidad y rumores de una oferta pública de acciones alimentan la desconfianza sobre si la empresa priorizará la protección del usuario sobre sus propios intereses comerciales.
- La función ofrece utilidad real —identificar gastos innecesarios, planificar inversiones, consultar con asesores fiscales— pero el precio implícito es entregar a una corporación tecnológica información que los propios bancos consideran su activo más sensible.
OpenAI ha lanzado una función que permite a los suscriptores de ChatGPT Pro en Estados Unidos vincular sus cuentas bancarias al chatbot, dándole acceso a historiales de transacciones, saldos, inversiones y deudas de más de 12.000 instituciones financieras. Con esta información, ChatGPT puede analizar patrones de gasto, sugerir estrategias de ahorro y orientar decisiones de inversión. La compañía también anunció una alianza con Intuit que permite a los usuarios agendar consultas con asesores fiscales directamente desde la plataforma, combinando análisis automatizado con orientación profesional.
OpenAI asegura que la conexión es cifrada, que el sistema no puede ver números de cuenta completos ni realizar modificaciones, y que solo accede a la información necesaria para responder preguntas financieras. Sin embargo, existe una trampa importante: de forma predeterminada, los usuarios tienen activada la opción que permite a la empresa usar los datos compartidos —incluida la información financiera— para entrenar versiones futuras de sus modelos. Desactivarla es posible, pero la compañía no ha detallado con claridad qué otros usos podría dar a esos datos.
Los expertos en seguridad no tardaron en encender las alarmas. Gang Wang, profesor de ciencias de la computación en la Universidad de Illinois, señaló que concentrar datos financieros en un sistema de IA introduce vulnerabilidades nuevas: una brecha de seguridad podría exponer no solo transacciones, sino patrones de comportamiento que los delincuentes podrían explotar con precisión quirúrgica. La información financiera es de las más cotizadas en el mercado negro digital.
El historial de OpenAI tampoco ayuda a calmar las dudas. La empresa ha modificado sus políticas de datos en múltiples ocasiones y ha enfrentado críticas por falta de transparencia. Las presiones para mejorar su rentabilidad y los rumores sobre una posible salida a bolsa han llevado a muchos observadores a preguntarse si sus promesas de privacidad son realmente sólidas.
La función representa un momento bisagra en la integración de la inteligencia artificial en la vida financiera cotidiana. La utilidad es genuina, pero el costo implícito —ceder a una empresa tecnológica el tipo de información que los bancos consideran su activo más valioso— exige una reflexión cuidadosa antes de pulsar el botón de conectar.
OpenAI has rolled out a new feature that lets ChatGPT Pro subscribers in the United States link their bank accounts directly to the chatbot. The move connects users to more than 12,000 financial institutions, allowing the AI to see transaction histories, account balances, investment holdings, and debt information. Once linked, ChatGPT can answer questions about spending patterns, suggest savings strategies, help plan investments, and discuss mortgages and other financial obligations. The company frames this as a way to give users personalized financial guidance without leaving the platform.
The feature includes a partnership with Intuit, the financial software company, that lets users schedule consultations with local tax professionals through ChatGPT itself. OpenAI says this combination of automated analysis and expert advice creates a more complete financial planning tool. The company reports that over 200 million people use ChatGPT monthly for help with budgeting, investment strategies, and financial planning—a large audience that the bank integration is designed to serve more effectively.
But the arrangement has triggered immediate concerns about data security and privacy. OpenAI states that the connection is encrypted and that ChatGPT cannot see full account numbers or make any changes to accounts. The system accesses only the information needed to answer financial questions: balances, transactions, investments, and liabilities. Yet there is a significant caveat. By default, users have enabled a setting called "Improve the model for all," which allows OpenAI to use the data they share—including their financial information—to train future versions of its AI models. Users can turn this off, but the company has not detailed what other uses it might make of the financial data collected.
Security researchers have raised alarms. Gang Wang, a computer science professor at the University of Illinois, told CNN that handing financial information to an AI system introduces new vulnerabilities. A data breach, security flaw, or cyberattack could expose sensitive transaction details that criminals could weaponize for phishing schemes or fraud tailored to individual users' spending habits. The risk is not theoretical: financial data is among the most valuable information on the dark web, and attackers have shown they will go to great lengths to obtain it.
OpenAI's track record on data handling has not helped ease skepticism. The company has changed its data policies multiple times and faced criticism for lack of transparency about how it uses information. CEO Sam Altman has been involved in public legal disputes, and the company has faced pressure to improve profitability—circumstances that have led observers to question whether OpenAI's privacy promises can be trusted. Rumors of a potential public stock offering have added to concerns that the company might prioritize revenue over user protection.
For users considering whether to connect their bank accounts, the calculus is complicated. The feature does offer genuine utility: having an AI that understands your full financial picture could help you spot wasteful spending or identify opportunities you might otherwise miss. But that utility comes with real risks. A breach of OpenAI's systems could expose not just transaction data but patterns of behavior that reveal where you shop, what you buy, and how much you earn. The default setting that feeds your financial data into AI training means your information is being used to build products that OpenAI will sell or license to others.
The feature represents a significant moment in how artificial intelligence is being integrated into everyday financial life. It is also a test of whether users will trust a technology company with information that banks themselves guard as their most sensitive asset. The answer will likely depend on whether OpenAI can demonstrate, over time, that it takes that trust seriously.
Notable Quotes
Sharing financial information with AI systems introduces new vulnerabilities that could expose sensitive data to phishing and fraud— Gang Wang, computer science professor, University of Illinois
The Hearth Conversation Another angle on the story
Why would someone actually want to give ChatGPT access to their bank account? What's the real benefit here?
The appeal is convenience and personalization. Instead of logging into your bank, then a budgeting app, then maybe a tax software, you ask ChatGPT one question and it has the full context. It can see you spent $800 on groceries last month and suggest where to cut back. It knows your mortgage balance and can model out refinancing scenarios. For someone overwhelmed by financial decisions, that's genuinely useful.
But that data has to go somewhere. What happens to it?
That's the uncomfortable part. OpenAI says it doesn't store your full account numbers and can't make transfers. But the data you share trains their models—unless you manually disable that setting. Most people won't know to do that, or won't bother. So your financial life becomes part of the training data for the next version of ChatGPT.
And if someone hacks OpenAI?
Then your transaction history, your spending patterns, maybe your income level—all of it is exposed. A criminal could use that to craft a phishing email that feels personal, or target you for fraud. Financial data is worth more on the black market than almost anything else.
Has OpenAI had security problems before?
Not major breaches that I'm aware of. But the company has been sloppy with privacy policies—changing them repeatedly, not being transparent about what they do with data. That's not the same as a hack, but it's a reason to be cautious about trusting them with something this sensitive.
So who should use this feature?
Someone who understands the risks, has read the privacy settings carefully, and has disabled data training. Someone who values the convenience enough to accept that their financial information is now in the hands of a private AI company. Not everyone should do this.