“Don’t stress about it! Our company only provides AI development as a service. We don’t fall under the AI Act scope.”
Businesses wear multiple hats
The same company can play multiple roles across the value chain and an evaluation from multiple perspectives is necessary before dismissing the compliance need.
The company can be a user of AI systems, encouraging employees to leverage existing solutions to more efficiently solve their tasks, which under the AI Act would make it a deployer.
The same company can develop internal tools with AI components, the results of which are impacting their employees, making the company both a provider and a deployer under the AI Act.
The company can develop an AI system and place it on the market, under its own trademark or name, free of charge or paid, like a chatbot for answering FAQ integrated on their website making the company a provider under the AI Act.
Service providers as trusted partners
Another hat can be that of a service provider which develops and rolls-out AI systems for their clients, without putting their own name on it, but acting as an external team of the beneficiary of the AI system. This doesn’t result in any legal obligations under the AI Act for the service provider. But depending on the risk level of that AI system, it might be impossible for the customer to use it in lack of supporting documentation, appropriate data governance measures, transparency, logging mechanisms to allow for continuous monitoring and much more.
AI Act compliance equals AI quality assurance
Complying with the AI Act as a service provider is a quality assurance requirement. The AI Act redefines how a high-quality AI solution should look like and sets the legal framework for placing reliable and robust systems on the market.
The companies who first understood the importance of aligning to the market change, as a competitive edge and a trusted partner, are taking steps towards collaborating with responsible AI leaders, creating internal audit teams, organizing AI Act webinars, equipping project managers and product owners with the relevant knowledge on how to build compliant AI solutions. Everything for the sake of building trust and reliability with their business partners.
New budgeting models
Justifying the nonconformity given the lack of legal binding is… to say the least reckless, moreover in a competitive and dynamic market. The notion of “quality of an AI system” is undergoing transformations, shifting the focus from pure model performance to trustworthy partnerships and compliant AI solutions.
Make sure to use that in the next sales pitch and budget all the additional work to meet compliance requirements.
Get in touch for:
– an on-demand training on the implications of the EU AI Act for your business or,
– a personalized AI Literacy course.
Reach us at: ioanang@giraffa-analytics.com