AI has the potential to revolutionize our world.
It also has the potential to cause harm.
Unchecked, AI can manipulate, exploit, and invade privacy. It can hurt individuals. It can harm society.
That’s why it’s great to see the European Union take a stand to regulate these dangers with the AI Act, which went into effect this month.
In this post, I’d like to focus not on the entire regulation, which is long and complex, but on one aspect: Article 5, which discusses prohibited uses of AI.
Let’s look at what the AI Act (Article 5) is doing to protect us:
- Subliminal and Manipulative AI: AI that uses subliminal or manipulative techniques to distort behavior and decision-making is banned. This prevents companies from using AI to trick us into actions we wouldn’t otherwise take, protecting our autonomy.
- Exploiting Vulnerabilities: AI systems that exploit vulnerabilities based on age, disability, or socio-economic status are prohibited. This means no more taking advantage of those already in vulnerable positions.
- Social Scoring: The Act forbids AI systems that evaluate or classify us based on our social behavior in a way that leads to unfair treatment. No one should be judged and treated unfairly based on an algorithm’s assessment of their social interactions.
- Predictive Criminal Risk: AI systems that profile individuals to predict criminal behavior are not allowed unless they support human judgment with objective evidence. This helps prevent biased and potentially discriminatory practices.
- Unauthorized Facial Recognition: Creating or expanding facial recognition databases through untargeted image scraping is banned. This protects our privacy and prevents mass surveillance without consent.
- Emotion Inference: AI systems that infer emotions in workplaces and educational institutions are prohibited, except for medical or safety reasons. This ensures our emotional states aren’t unfairly monitored or used against us.
- Biometric Categorization: Categorizing individuals based on biometric data to deduce sensitive personal attributes is restricted, safeguarding our personal information from misuse.
- Real-time Remote Biometric Identification: The use of these systems in public spaces for law enforcement is tightly controlled and requires proper authorization, preventing misuse and protecting our rights in public spaces.
The AI Act is controversial, with its opponents contending that it harms innovation. However, in my view, these eight prohibited uses seem pretty universally agreed upon as unethical.
What’s really important is that the AI Act has a broad scope and strong extraterritorial reach. It applies to any AI system impacting the EU market, regardless of where the provider is located. So, it covers AI systems placed on the market, put into service, used by an EU-based user, or when the output is used within the EU.
Despite the criticism, I think we should applaud the EU for setting a global benchmark for responsible AI governance.
This is a significant step towards ensuring AI serves humanity positively and not as a tool for manipulation and enslavement.
