top of page
Writer's pictureGreen Cabbage

AI’s Sword of Damocles: Aggressive Strategies for Risk Control



In today’s AI-driven world, the control you think you have over your data might be as tenuous as the thread suspending the Sword of Damocles. As AI systems advance, the complexities and risks of data management intensify, threatening to snap that thread at any moment. Picture this: the technology suppliers you trust may be using your data to fuel their AI, commingling it with other customer's data, or worse - using your intellectual property as training data to enhance their services, all without your explicit knowledge. It's time to take control. This white paper isn't just another guide - it's your blueprint for aggressive data ownership strategies, arming you with the knowledge to lock down AI input and output rights, and to manage the hidden dangers of AI technology. From security and privacy breaches to the complex web of third-party liabilities, we're covering it all.


Data Rights: Asserting Control

Data ownership, especially with intangible assets like information, is legally complex. As a customer, your goal is to assert your intellectual property rights affirmatively while ensuring your data is used in ways that align with your strategic interests. Your contract must clearly define how your data can be used, who controls it, and what limitations are in place. Traditional nondisclosure agreements do not fully capture the intricacies of AI data use, so it is critical to negotiate a master agreement that explicitly governs data usage and confidentiality throughout the AI lifecycle. Ensure that your contracts leave no room for ambiguity, particularly when dealing with AI systems that rely on your prompts, training data, or input data.


Nondisclosure for Prompts and Customer Data

When integrating sensitive data, such as prompts and training data, into AI systems, it is essential to lock down this information as "Confidential Information" within your nondisclosure agreements. Insist on including exceptions only for independently developed information, ensuring your data stays securely within your control. Nondisclosure agreements should also be tailored to cover AI-specific risks and ensure data remains under your control. Best practices include data anonymization, obtaining explicit consent, and implementing rigorous monitoring systems.


Challenges with AI Systems Training for Broader Use

AI systems that utilize your data for multiple customers introduce a level of complexity that demands your attention. By negotiating clear boundaries on data usage and considering alternative protective measures, customers can work collaboratively with suppliers to ensure their data remains secure and exclusive. However, be cautious about providing sensitive information that could be repurposed beyond your specific instance of the AI system. If the supplier resists strict nondisclosure terms, push harder – your data, your rules. The goal is to keep your data exclusive and under your control.


Protecting Inputs, Outputs and Trade Secrets

Extend the protections of your nondisclosure agreements (NDAs) to include AI Inputs and AI-generated outputs. Treat every input and output as a potential trade secret and maintain confidentiality until you have thoroughly assessed its sensitivity. Do not let the supplier’s system dictate the handling of your data, inputs, and outputs - assert control from the outset. While asserting control is important, it's equally vital to consider broader IP protections, such as securing patents or copyrights where applicable.


Challenging Supplier Resistance

Suppliers will likely push back, especially if the outputs include data or content owned by third parties or the supplier itself. Negotiating effectively requires balancing assertiveness with an understanding of the supplier's perspective. Push for explicit terms in your favor and be wary of intellectual property disclaimers designed to minimize their liability. Your contracts should secure ownership and anticipate any strategies suppliers might use to limit your rights.


Challenging Supplier Disclaimers

Never accept supplier disclaimers without thorough scrutiny. If AI outputs involve third-party content provided by the supplier, demand IP and privacy warranties and indemnities. The complexity of AI systems may make it difficult to trace the origin of specific output elements, but that’s more reason to negotiate detailed contractual terms.


Conclusion

The risks of AI technology may continue to precariously hang over your business - offering the potential to elevate your success but always threatening to fall with hidden liabilities if not carefully managed. As you navigate this intersection of AI technology, intellectual property rights, and security, remember that passivity is your greatest enemy. The strategies outlined in this paper are not just suggestions; they are essential actions you must take to shield your business from the lurking threats that could undermine your profitability. Every contractual clause and every safeguard must be optimized to protect your interests. Secure those robust warranties, indemnities, and data protection agreements. Your data is your power - never let it be wielded by a third party whose only interest is their own earnings report. The time to act is now; the consequences of inaction are far too great.


Are you ready to begin with a team of experts by your side? Click here to request a demo!


 Written by: Wesley Schnell Chief Legal Officer, at Green Cabbage

Comments


bottom of page