While many organizations are looking at ways to incorporate LLMs, they also pose numerous challenges around data privacy. Data security company Symmetry Systems is helping companies tackle these challenges so that organizations can safely consume AI technology like LLMs. Claude Mandy, Chief Evangelist for Data Security, Symmetry Systems, says that organizations “don’t necessarily know what the model was trained on. Or they don’t have control of visibility of what’s in that data that they’re training these new LLM models on.”

AI and data privacy challenges in the enterprise space

• Symmetry Systems is a data security company that provides visibility and the ability to remediate the issues, and protection for organizations’ data.
• Mandy discusses the problem of organizations not being able to answer questions about their data, including who has access to it and what they’re doing with it.
• Mandy explains the increasing adoption of data privacy regulations akin to GDPR, emphasizing the importance of data sovereignty and residency, particularly in ensuring citizens’ rights over their data across borders.

Data privacy challenges in AI development and use

• While LLMs can enable organizations to go quicker, there are challenges around privacy and data because many organizations do not know what the models were trained on and they do not have visibility or control over what is in the data.
• Mandy highlights the need for organizations to be privacy-conscious when using LLMs, as they may unintentionally reveal sensitive information through prompts and output.
• Mandy discusses how organizations are responding to data privacy concerns around AI, such as by blocking access to them until they have the right controls on it or investing in educational efforts to educate users on responsible use.
• Mandy believes companies must focus on creating data inventories and AI inventories to implement controls to restrict access to personal data used to train AI models.

AI ethics and regulation

• Mandy describes the measures companies are taking to restrict AI access and the risks associated with it. He discusses the challenges of new technologies and how it takes time for legislation to catch up.
• Individual companies as well as the industry as a whole and lawmakers need to work together to address issues like data privacy and ownership.
• Mandy emphasizes the importance of ethical data use and proper governance through legislation to ensure responsible AI development and adoption.

AI adoption in enterprise, focusing on guardrails and risk management

• Mandy discusses the importance of having the right tools and guardrails so that developers know what data is in what environments and go through the appropriate steps to use that data, which starts with visibility and classifying it.
• Enterprises need to be able to remediate if data ends up where it is not meant to by reducing the amount of personal data developers have access to, and monitoring the LLMs with proactive and response tools.
• Mandy explains how Symmetry Systems’ tools and hands-on support are helping organizations use AI tools safely by developing classification processes, identifying important data, and putting guardrails in place to mitigate risk.

AI adoption in the industry: risks and opportunities

• Venture capitalists are investing heavily in AI; however, enterprises are being more cautious in adoption, looking for the right use cases.
• Mandy believes we are still seeing AI being overhyped but in the long term, it will create a massive opportunity for the whole industry. He explains how Symmetry works with organizations to work through these processes.

Guest: Claude Mandy (LinkedIn)
Company: Symmetry Systems (Twitter)
Show: Let’s Talk

This summary was written by Emily Nicholls.