[ad_1]
A 37-hour-long discussion has led to the world’s first set of AI laws. A historic agreement to regulate the unchecked growth of artificial intelligence was signed between the European Parliament and EU member states.
The AIAct is much more than a rulebook — it’s a launchpad for EU startups and researchers to lead the global AI race.Thierry Breton, European Commissioner
The agreement is based on a risk-based tiered system, which would impose more liability on AI models that bear the highest risk.
The level of risk is determined by the training process of the AI model — the higher the number of computer transactions needed, the higher will be the risk. This metric is named floating point operations per second (Flops). Currently, only ChatGPT 4 satisfies this condition.
However, the agreement will include all systems with 10,000 business users.
Ban On Real-Time Surveillance
European Commissioner Thierry Breton expressed his joy over the newly signed AIAct, mentioning that it was “worth the few hours of sleep”.
The Act is not expected to come into force late last until early 2025.
One crucial decision reached was a ban on the use of AI for real-time surveillance and biometric tech by employers, police, and other organizations. However, there are three exceptions.
The police will be allowed to use AI surveillance in case of a foreseeable threat, such as terrorist attacks, to search for victims and prosecute serious crime offenders.
An understanding was reached that police would need prior permission from “independent authorities” before they can employ “predictive policing”.
Dragoș Tudorache, the Romanian MEP, assured that the house was not against the use of essential police tools to fight and combat crimes in Europe. However, the use of AI to predict who would commit a crime was unethical and hence, a ban on the same was imposed.
The Act has also imposed strict penalties on violations:
- For violation of banned AI apps — €35 million or 7% of the Global Annual Turnover (GAT) of the violating company in the previous year.
- For violating Act provisions — €15 million or 3% of the GAT of the violating company in the previous year.
- For supplying false information — €7,5 million or 1.5% of GAT
The AI Race
This move puts Europe ahead of major competitors like USA and China, who have also been trying to regulate AI.
Anu Bradford, an EU and digital regulation expert, believes that this move has set a precedent for other regulatory bodies to follow. Many other regulations from around the world will likely borrow provisions from the EU AIAct.
Tudorache also shed light on the promptness of the decision. He highlighted the need to learn from past mistakes when giants like Facebook were allowed to operate without any restrictions. This has led to concerns like child abuse, unregulated content, and election manipulations.
The EU was in no mood to repeat these mistakes with current AI giants. Hence, the regulation imposes liability on AI companies to develop a human-centric approach.
Business Houses Criticize The Act
As expected, prominent businesses have criticized the new AIAct, terming it as yet another burden for companies.
We fully supported a risk-based approach based on the uses of AI, not the technology itself, but the last-minute attempt to regulate foundation models has turned this on its head.Director General Cecilia Bonefeld-Dahl, DigitalEurope
Whatever the reactions might be, the EU seems to be firm on its decision to control the unchecked intrusion of AI into people’s privacy. The path to execution is expected to be bumpy.
[ad_2]
