In the realm of artificial intelligence, the emergence of public generative AI tools has become a focal point of interest in recent years, leading to various applications incorporating new chat capabilities and other features driven by large language models. However, organizations are facing a new challenge as they endeavor to connect these Language Model (LLMs) tools to their internal knowledge repositories, with potential risks lingering in the background.
One of the core issues that have surfaced with the integration of LLMs into internal systems is the unintentional exposure of sensitive information. Sounil Yu, co-founder and CTO of Knostic, highlights this concern as he notes that business leaders have been taken aback by the tools providing answers to queries such as ‘What are people’s salaries?’ and ‘What are the most recent M&A due diligence results?’ This inadvertent access to confidential data can pose a significant threat to organizational security, even if the permissions and access controls are technically in place.
Gadi Evron, co-founder and CEO of Knostic, echoes this sentiment by underscoring the increased risk that AI without proper controls can bring. He emphasizes the importance of delivering personalized and valuable information, tailored to the user’s specific needs, rather than indiscriminately disclosing vast amounts of data. In an effort to address these challenges, Knostic has emerged as a pioneering force in defining per-user need-to-know parameters and establishing a robust knowledge control layer within organizations.
Unlike conventional approaches that focus solely on data scanning, permissions, and information classification to tackle oversharing, Knostic’s technology offers a more comprehensive solution. By providing visibility, control, and curation capabilities, the platform enables organizations to query genAI tools for sensitive topics, evaluate permissions for content, and customize access levels based on user requirements. This nuanced approach ensures that users only receive information that is relevant to their roles and responsibilities, mitigating the risks associated with oversharing.
Furthermore, Knostic’s technology advocates for a more nuanced approach to access control, rejecting the binary nature of traditional permission settings. Rather than a simple ‘yes’ or ‘no’ approach, the platform empowers security teams to curate search query responses based on the user’s need-to-know level, promoting a more tailored and secure information-sharing environment.
Initially focused on integrating with Microsoft’s Copilot within M365 environments, Knostic is now expanding its reach to address need-to-know challenges across a broader range of software-as-a-service tools featuring LLM capabilities. This strategic expansion underscores the company’s commitment to enhancing knowledge sharing practices and safeguarding sensitive data across various technology platforms.
As a testament to its innovative approach, Knostic has recently been recognized as a finalist in the prestigious Black Hat Startup Spotlight competition. Alongside other pioneering startups in the cybersecurity space, Knostic will showcase its business model and technology to a panel of industry experts at the Black Hat USA Conference in Las Vegas. This unique opportunity to demonstrate their solution underscores Knostic’s commitment to addressing the evolving challenges of data privacy and information security in today’s digital landscape.
In conclusion, Knostic’s proactive stance on redefining knowledge control and mitigating the risks associated with oversharing represents a significant advancement in the realm of AI-driven knowledge management. By prioritizing personalized access, visibility, and control, Knostic is spearheading a new era of secure information sharing that aligns with the evolving needs of modern organizations.