In a recent webinar forming part of DLA Piper’s ‘Digital Evolution in conversation with’ series, Technology Transactions and Sourcing partner Lauren Hurcombe caught up with Gareth Stokes and Jeanne Dauzier, DLA Piper’s International Co-Chairs of the AI Practice, to discuss the opportunities and challenges posed by navigating AI from a sourcing perspective.
Contracting for AI
Governments and regulators are currently grappling with the desire to regulate and set restrictions on the implementation and use of AI, while at the same time ensuring that they do not stifle innovation. We also see businesses and their in-house counsel, C-suite and technology teams trying to manage the impact of AI on the procurement of solutions and the negotiation of the underlying technology contracts. This presents a need to adapt contracts to address the unique risks and realities of AI solutions.
It is crucial to understand the technology stack and contracting model to assess risks appropriately, as contracting for AI systems differs from other technologies and requires input from operational and business teams, not just legal teams. Ensuring that sales, security and procurement teams are aware of how AI systems operate, and associated risks, is also essential. Using an AI assessment questionnaire and asking the right questions upfront to avoid last-minute surprises can be a valuable part of the procurement process.
Impact of the changing AI regulatory landscape on contracts
Different regions have varying regulatory approaches to AI and these regulations are still evolving. The introduction of the EU AI Act heralded the first comprehensive legislation addressing AI. Various regions, including China, the US, and Japan, are developing their own AI regulations, with some states in the US already implementing these. Increasing regulation has caused market nervousness, leading some tech companies to delay AI product launches in Europe. Despite this, businesses must still comply with EU regulations due to their extraterritorial impact.
It is important to consider upcoming legislative developments and their impact on contract negotiations. Existing SaaS contracts often need modifications to definitions and clauses to address AI-specific elements like input, output, and training data. AI contracts may need to include new clauses for aspects such as bias, discrimination and ethical considerations, which are not typically found in standard SaaS agreements. With AI solutions, especially generative AI systems, there is limited room for negotiation. It is more efficient to assess the risks and decide if the solution fits the business needs.
Applying a pragmatic and flexible approach during contract negotiation, seeking to balance interests between AI customers and service providers, can be conducive to reaching agreement. The parties need to discuss and understand the reality of risks to find a balanced approach in contracts, considering financial and reputational risks. The compliance with laws clause in contracts will undoubtedly gain more attention going forward due to the evolving legal framework and financial implications of AI regulations.
The impact of AI on sourcing models and infrastructure
Due to the traditional offshoring models that are already in place onto which AI systems may be superimposed, the challenge of having a long supply chain rears its head, with people often feeling more comfortable working with local infrastructure. We have seen this with the cloud, where there has been a build up of lots of different infrastructure, data centres and so on, within different markets. The idea of having a locally hosted cloud solution to provide comfort of compliance with rules around cross border data transfers has driven changes in the market, and it is likely that similar factors will drive changes in the location of service delivery as we move towards fuller AI adoption.
The fact that AI is a big part of service delivery does not necessarily mean that we will see a big swing away from offshoring as a solution, however, because a lot of activities are mixed in nature. Even if AI can do half of the tasks the e.g. a human powered delivery centre is doing, that means we are still left with a potentially large number of humans on site.
We are starting to see increasingly tough export control rules, particularly around the most cutting-edge semiconductor products. The chips that are powering AI are increasingly difficult to get hold of outside Western jurisdictions, making it much easier for people to set up big data centres in the US, for example. The availability of raw materials will impact on the concentration of data centres in certain geographies, at least in the medium term. This combination of factors may mean that we see a push towards reshoring of some services, particularly the AI powered elements of those services, and a mixed delivery model, with onshore/offshore being the default model in a in an AI powered world.
Looking ahead
AI’s boom may well lead to shorter, more flexible sourcing contracts, impacting negotiating positions and the ease of switching solutions. The commercial models for outsourcing will change significantly due to AI, affecting how services are priced and delivered. The availability of skilled professionals such as data scientists may also pose a significant impact, and one that urgently needs to be addressed in many regions.
A one-size-fits-all approach is impractical due to regional regulatory and technological differences, necessitating flexible and localised solutions. There is a need for a holistic but adaptable AI procurement process that accounts for regional nuances and customer demands. The importance of vendors understanding and adapting to changing regulations and customer needs will be a key factor going forwards.
[View source.]