You’ve likely heard the saying that “data is the new oil.”
It is particularly true in our age of artificial intelligence (AI). But throughout history, wars have been fought over oil — and today, data is no less a critical resource to protect and secure.
That’s because AI has changed to the way businesses approach data transformation, storage, management and governance.
This is an unavoidable reality within security-critical sectors like payments and financial services, where conversations around data transformation, data sharing and generative AI are top of mind for every executive — as are the implications of AI on data collection, storage and use.
“For the last 10 years, the financial sector alone has invested countless sums into data science and building out their data intelligence departments,” Taylor Lowe, CEO and co-founder of AI infrastructure platform Metal, told PYMNTS.
“Not only is a lot of the infrastructure in place throughout these organizations, but the incentives are there: the years of operating with a data-first mindset have paid off, and firms see how valuable it is. Technologies like AI have just added more fuel to that fire and accelerated the insights you can get out of your data,” Lowe explained.
But the increasing value of data has not occurred in a vacuum, and as we explore in the “Attack Vectors 2024” series, today’s data landscape doesn’t just create opportunities for organizations, it also opens up vulnerabilities and attack vectors that need to be prioritized.
Read more: Why Every Business Now Wants a Data Lakehouse
Companies often store intellectual property, trade secrets, and proprietary information in their data. Unauthorized access to this data can be especially damaging.
As organizations looking to deploy AI applications across their internal workflows will need to tag, restructure, transfer and migrate their own data — meaning that the security of that information should be of the utmost priority.
Lowe noted that at the chief information security officer level, existing security practices such as encryption and network segmentation still apply when undertaking an AI deployment, but organizations may need to adapt their internal programs to better meet the needs of evolving AI systems.
“We are still in the early innings, and data migrations don’t happen overnight,” he said. “Secure options need to be considered, like deploying AI applications on your own network, which is something Metal provides.”
After all, you need guardrails if you are going to ask an organization to entrust its most valuable asset, its own data, to an AI application.
Still, enterprises generate and store vast amounts of data in various formats and locations, and an unavoidable reality is that integrating this data into a coherent and accessible form is frequently a challenge.
But it is one worth confronting. The power of generative AI tools to spread through every enterprise function and support every employee, as well as to engage every customer, rests above all on data.
Read also: Companies Tap Their Own Data to Drive Efficiencies With AI
Lowe emphasized the importance of harnessing unstructured internal data, which accounts for about 80% to 90% of the world’s data. With the introduction of AI and large language models (LLMs), unstructured data can be made much more useful and valuable.
“LLMs can read through unstructured data with amazing results, but they still need direction. The insights you’re after will inform the use cases for your data — which is what the software you’ll use needs to be built around,” he explained.
Effective execution depends on two parallel tracks. The first is infrastructure to support the transformation, storage and querying of existing data; while the second is software that uses this infrastructure in service of specific workflows.
“Marrying these two is where you’ll see real productivity gains,” said Lowe.
Of course, if a company is taking sensitive data, transforming it, and storing it — tracing that data’s lineage in case of an audit is critical, as it protecting it in real-time.
And when it comes to deployment, Lowe underscored that “humans absolutely need to be a part of the workflow and system. … There’s an incredibly high bar for accuracy in areas like financial services, and there needs to be the ability to override and correct outputs.”
“Today’s AI are best thought of assistants that need supervision — especially in the world of finance,” he added.
By adopting a comprehensive approach to data security, companies can better safeguard their data and mitigate the potential vulnerabilities associated with the use of AI.