TLDR:
- Banks are turning to hyperscalers for cloud computing and generative AI solutions
- Concerns arise about concentration risk, outages, cybersecurity, and data privacy regulations
First, banks have migrated software applications to clouds hosted by Amazon, Google, and Microsoft. Now, they are utilizing large language models from these same companies, known as hyperscalers. Some worry about concentration risk with so much of the financial system relying on the servers of a few tech giants. Concerns include the potential for outages, cybersecurity threats, price increases, and adherence to data privacy regulations.
As banks experiment with large language models, their vendor shortlists often feature companies like Google, Microsoft, Amazon, IBM, and OpenAI. The Treasury Department has expressed concerns about cloud providers’ lack of transparency, cybersecurity expertise shortages within banks, and potential cascading effects of cyber incidents.
There is a risk of multiple banks being misled by errors, hallucinations, or bias in large language models. Some experts suggest that open-source communities like Hugging Face offer hope for innovation and diversity in AI models. However, the concentration of advanced AI models in a few companies remains a concern.
Ways to mitigate concentration risk include working with multiple cloud providers and embracing hybrid cloud strategies. Governments may need to intervene with regulations to ensure these technologies are used responsibly. The responsibility for addressing these risks lies with elected representatives, according to some experts.
In conclusion, the increasing reliance on AI and cloud computing in the financial sector poses risks related to concentration, cybersecurity, and data privacy. Collaborating with multiple providers, embracing open-source innovation, and implementing regulatory oversight may help mitigate these risks in the future.