Thinking Space

Proposed EU AI act and it’s impact

The proposed EU AI act which is currently in its final stages is going to be world’s first comprehensive legal framework for AI. Technology leaders must be prepared to put a plan to deal with AI risks, trust and security management.

Key elements of the current draft include:

  • The act will apply to any AI systems placed in the EU market or affecting EU citizens. So, similar to GDPR it will affect the organizations outside EU as well.
  • AI use will be regulated based on their level of risks.
  • Non-complaint organizations will have to pay heavy penalties.

It is has proposed three risk categories:

  • Unacceptable risk: AI systems clearly threatening fundamental rights will be banned. Some possible examples in this category will be biometric systems, emotion tracking at workplace etc..
  • High risk: AI systems with potential negative impact on people’s safety. Some possible examples in this category will be information of users, logging of activities.
  • Minimal risk: AI system with narrow and tactical objectives.

Most AI systems will be subject to additional transparency requirements. For examples users will have to be informed when talking to chatbots, or when a generated content is a ‘deepfake’.

Potential impacts:

Negative:

  • In short term it will create a barrier for EU companies and their competitiveness outside EU.
  • It could also become a barrier for AI adoption in Europe.

Positive:

  • In long term the compliant organizations will have start having competitive advantage.
  • By having a unified standard for responsible use it will create a trust and adoption with users.

Before making big investments to the AI/Gen AI solutions, be prepared for the AI act and assess the risk categories.

Dushyant's avatar

Posted by:

Dushyant

2 responses to “Proposed EU AI act and it’s impact”

  1. Too Much, Too Soon, Too Fast! – Thinking Space

    […] I think something similar is happening with AI, we started with LLMs such as Open AI’s ChatGPT. Google Bard(now called Gemini) followed soon. We now have Open AI’s Sora which can generate great quality videos with text prompts. These are powerful tools and have some valid use cases, but, many organizations are rushing to fit their use cases with the capability of tools rather than the real business benefits. As I wrote in my earlier blog ‘Generative AI – is your organization ready?’, organizations need to first focus on being AI ready with AI ready security, AI ready data and AI ready principles and think about the readiness to be compliant of the upcoming AI regulations such as ‘EU AI act’. […]

    Like

  2. AI Trust, Risk & Security Management – Thinking Space

    […] AI TRiSM(Trust, Risk and Security Management) is a framework that comprises a set of risk and security controls and trust enablers that helps enterprises govern and manage AI models and AI applications’ life cycle. It will also help organizations comply with upcoming regulations, such as the EUs AI Act. […]

    Like