Navigating AI: Mastering the Balance of Innovation and Compliance
Expert Insights by Sabine Reifenberger, Frankfurt
AI regulations and limitations for safe, responsible use
When Europe greenlit the General Data Protection Regulation in 2016, legal eagles were kept busy. Now, it's AI that's causing a buzz, with Christoph Werkmeister, global co-head of the data and technology practice at Freshfields, pointing out that AI influences industries beyond tech: retail, auto, and health are hopping on the AI bandwagon. However, leaders need to tread carefully to avoid chaotic outcomes and epic fails.
The EU's AI Regulation, which went live in August, offers a roadmap for AI governance within companies, says marathon runner Klaus Brisch, partner at Grant Thornton. The regulation takes a risk-based approach, with higher demands placed on high-risk systems, such as those using biometric data or impacting safety-critical areas like traffic and energy supply. These systems require intensive testing for compliance via external audits, a process known as conformity assessment.
The onus is on boards and CEOs to practice due care, according to Müller-ter Jung, another partner at Grant Thornton. This means opting for certified and well-documented AI systems, cutting down on mistakes. Companies must also be ready to provide regulators with updates about their AI systems' functionality.
With responsibility comes the need for workforce AI literacy. While the EU AI Regulation stops short of providing a detailed training protocol, it does demand that companies train their employees on AI when they implement the technology. Werkmeister of Freshfields advises viewing training as more than a regulatory chore. Comprehensive knowledge of AI in-house enables better decision-making when dealing with providers and assessing suitable systems.
The journey to establishing an AI governance structure has no definitive blueprint. Werkmeister sees a parallel with data protection, suggesting that companies with experience in the area will find it easier to adapt. Müller-ter Jung concurs, urging companies to leverage their GDPR knowledge to streamline their AI governance efforts.
For global players, the regulatory landscape becomes a patchwork quilt. The EU regulation, although applicable to non-EU corporations making a splash on the European market or supplying AI outcomes in the EU, is just one piece of the puzzle. Different countries prioritize distinct regulatory angles, such as consumer protection or copyright.
The future is uncertain, with some countries considering more extensive AI governance measures. Werkmeister advocates a focus on the fundamentals - risk management, transparency, and documentation - and ongoing monitoring of AI behavior. Preparation is key to future-proofing AI strategies.
Essential AI Regulation Insights:
- Governance Structure: Establishing a governance structure that ensures transparency, accountability, and data protection is crucial for AI adoption.
- Risk-Based Approach: AI systems should be categorized according to their risk level to implement appropriate measures.
- High-Risk AI Systems: High-risk AI systems demand more rigorous compliance, including human oversight, data protection, and conformity assessments.
- Standardization: Adherence to approved codes of practice and industry standards showcases compliance and promotes consumer trust.
- Regulatory Bodies: The European AI Office monitors compliance, while national authorities oversee enforcement.
- Liability Framework: Legal liability rules surrounding AI-related damages are evolving, and companies should stay informed.
- To effectively navigate the AI landscape while maintaining compliance, companies must incorporate AI governance structures that embody transparency, accountability, and data protection.
- The EU AI Regulation adopts a risk-based approach, imposing stricter compliance measures on high-risk AI systems, such as those using biometric data or impacting safety-critical areas.
- Adherence to approved codes of practice and industry standards can help demonstrate compliance and build consumer trust in AI systems.
- With different countries having varying AI governance regulations, global players should stay informed about new developments and focus on fundamental aspects like risk management, transparency, and documentation to future-proof their AI strategies.
