• DINT
  • Posts
  • DINT 78 - The Final Hours for AI Laws in Europe - Will They Be Tough Enough?

DINT 78 - The Final Hours for AI Laws in Europe - Will They Be Tough Enough?

PLUS: NY Bans Facial Recognition in Schools

Soon a high-stakes part of the development of AI will reach its final stages. Here we share an overview of the key issues and explore what could happen next.

3/ Watch Out! Food Delivery Robots Send Data to LAPD: 404Media

5/ Notes on Linux from the Cloud Native Computing Foundation Conference aka Kubecon

Friday Feature

Racial Bias in AI Is a Human Rights Issue: Amnesty International

The European Union’s (EU) governing body is putting the finishing touches on its landmark AI legislation, and many organizations are working to influence the EU’s final draft.

It’s a high-stakes game that puts the regulation of racial, gender, and ethnic bias on the line, potentially costing Big Tech trillions of euros in fines if AI models violate the EU’s AI law.

According to Stanford University’s Human-Centered AI Center, there are three key areas of importance:

  • AI use for biometric surveillance in public spaces. Parliament introduced prohibitions that limit law enforcement’s use, which are likely to be objected to by member state governments in council.

  • Definition of high-risk AI. Parliament allows for a broader definition of use cases for high-risk AI, while the council would prefer to narrow that definition.

  • Governance. The three will need to discuss, among other things, national silos in terms of implementation and enforcement, as well as levels of coordination.

    (The above bulleted list is an excerpt from Analyzing the European Union AI Act: What Works, What Needs Improvement by Shana Lynch)

Will Human Rights Get the Same Protection as Data Privacy?

This is happening in a climate where Europe has some of the most stringent and expensive data protection laws in the world, levying more than two billion euros in penalties on Big Tech in the last five years.

If new AI legislation takes a similar shape as the EU’s data privacy regulations, Big Tech will have costly changes to make in their evolving AI business models.

Current AI Climate: Rampant Bias and Pushes for Self-Regulation

Among the groups working to influence the EU’s AI legislation is the global human rights organization Amnestly International. The organization calls for no self regulation for Big Tech organizations’ use and development of AI, something Big Tech is fiercly fighting against.

These systems are not used to improve people’s access to welfare, they are used to cut costs. And when you already have systemic racism and discrimination, these technologies amplify harms against marginalized communities at much greater scale and speed.

Mher Hakobyan, Amnesty International’s Advocacy Advisor on AI Regulation

Will AI Violations Cost Big Tech as Much as Data Privacy Misdeeds?

Since GDPR’s inception, Big Tech has paid 2 billion euros in fines (see image below). If the EU’s legislative arm imposes the same type of stringent (and actively enforced) regulations, AI could look quite different in the coming decades. Here are some stats from the organization’s database via the EQS Group:

Source: EQS Group

According to Amnesty International a fineable offense in EU AI legislation would be the Netherlands’ use of AI to deny state-sponsored benefits based on the race of potential recipients.

… an AI system used by the Dutch tax authorities had racially profiled recipients of childcare benefits. The tool was supposed to ascertain whether benefit claims were genuine or fraudulent, but the system wrongly penalized thousands of parents from low-income and immigrant backgrounds, plunging them into exorbitant debt and poverty.

Will AI Fines Be as Hefty as GDPR Fines Have Been?

According to EQS, the most common fine categories under GDPR are:

  1. 💶 1.2 billion noncompliance with data processing principles

  2. 450 million - insufficient legal basis for data processing

  3. 375 million - insufficient protection of data

  4. 237 million - lack of communication with affected parties of data violations

  5. 50 million - violations of data subjects’ rights

Source: EQS Group

The U.S. has yet to pass comprehensive legislation protecting citizens’ data. The California Consumer Privacy Act (CCPA) is the nation’s only answer to the gap in data privacy oversight.

What’s Next for EU’s AI Law?

In October, EU legislators will convene to finalize its AI laws with adoption of the laws to span three years into 2026.

What’s Happening in Tech

Black Tech Fest
October 10
Virtual

Palate Cleanser

It looks like this iPad user may need another one. It looks like her dog has taken this one over! 😆 

@carma634

I’m going to need a new ipad #jollydog #whathaveidone

Reply

or to participate.