Safeguarding Sensitive Data and Intellectual Property from Large Language Models? Introducing LLM Shield

Serial entrepreneur Wayne Chang launches LLM Shield which leverages AI to protect sensitive data from large language models like ChatGPT.

San Francisco, CA — April 20, 2023 — Patented.ai, the world’s first intellectual property software built from the ground up on AI-first principles, today announces its introductory tool, LLM Shield. LLM Shield allows companies to safeguard their most sensitive data – proprietary source code, private customer information, unreleased financial data, legal documents, board reports and more – from large language models (LLM), such as OpenAI’s ChatGPT, Anthropic’s Claude and Google’s BARD. As the most effective and efficient enterprise resource for protection and compliance when working with LLMs, Patented.ai’s LLM Shield is set to roll out additional data checks and product offerings throughout the year including a personal version soon.

The recent news of Samsung’s data leak to ChatGPT underscores the risks associated with using large language models, as it exposed proprietary source code and business strategy. In light of such security reasons, Bank of America, Goldman Sachs, Citigroup Inc., Deutsche Bank AG and Wells Fargo & Co. have all banned ChatGPT (Source) to minimize the chances of confidential and personal identifying information being leaked. While LLMs offer the ability to help professionals across industries, IP-forward companies remain concerned with accountability and liability. 

LLM Shield beta offers free sensitive data checks for companies and professionals in fields such as advertising, finance, IT/security, legal, marketing, military, sales as well as personal identifying information. From there, companies can utilize LLM Shield’s open-source digital security tool designed to prevent company employees from inadvertently sharing this sensitive data when working with LLMs. This minimizes chances of losing competitive advantage, reputational damage, financial losses and legal ramifications. To maximize security, the software is built to be hosted within the company’s existing IT infrastructure, inheriting all the existing benefits and security investments. With LLM Shield integrated, companies can deploy to their employees and select the level of sensitivity for the data – and the actions they want their employees to take. Options include a simple warning that pops up within ChatGPT, to code abstraction, obfuscation and data masking, to a hard stop on the most sensitive data, ensuring the security of the company’s sensitive data and intellectual property.

According to McKinsey, a majority of consumers value digital trust, yet few companies are actively mitigating digital risks. Designed to address the pressing need for data protection in the digital realm, LLM Shield offers:

• Trustworthy security – effectively shields company secrets and sensitive information, such as passwords and secret keys, without impacting language model performance.
• Multiple layers of protection – conceal unique parts of code by disguising them, remove any hidden messages within the code that might reveal secrets and simplify the code while preserving its function. 
• Adaptable design & flexible deployment – seamless integration, extensibility, and adaptability across settings, from simple Google Workspace Chrome plugins, to advanced VPN integrations.

Founded by serial entrepreneur and investor Wayne Chang, Patented.ai and LLM Shield significantly reduce the associated risks of commercial use of these cutting-edge technologies. This initial offering democratizes the ability for anyone to safeguard their sensitive information and source code, which has long been only feasible for major companies with large legal teams. 

“AI is incredibly powerful and companies shouldn’t have to ban access,” said Chang. “With visibility and control, companies should have more confidence with their employees and their use of LLMs.”

LLM Shield currently supports sensitive data across advertising, finance, IT/security, legal, marketing, military, sales and personal identifying information, with more to come, as well as source code monitoring. Employee deployment can be done through its Google Chrome Plugin – Google Workspace customers can deploy it to all their managed users instantly. Support for other deployments, such as VPN and Proxy support, are also in the works. 

Patented.ai is committed to creating a seamless and efficient experience that empowers inventors, businesses and legal professionals to protect IP, paving the way for a more accessible approach to solving the problems of the intellectual property ecosystem by setting the industry standard.

For more information about LLM Shield, visit https://llmshield.com.

About Patented.ai

Patented.ai is a pioneering intellectual property engine on the frontlines of safeguarding and revolutionizing the way we protect IP. As part of the Chang Corporation, founded by Wayne Chang, a serial entrepreneur with a diverse portfolio of successful ventures, Patented.ai exemplifies Chang’s commitment to empowering innovators and businesses in a rapidly evolving digital landscape. Its flagship product, LLM Shield, marks the beginning of the company’s mission to redefine the industry standard for IP protection, while significantly reducing the time, investment and barriers associated with safeguarding invaluable ideas and hard work. Driven by the belief that the essence of progress is the protection of creativity and innovation, Patented.ai strives to create a world where intellectual property is secure, fostering an ecosystem in which brilliant minds can confidently build, invent and contribute to the greater good, today and into the future.

Learn more at LLM Shield.

Media Contact: patented.ai@astrskpr.com