Today’s Cache | EU’s AI Rules get lawmaker approval
| Photo Credit: REUTERS

(This article is part of Today’s Cache, The Hindu’s newsletter on emerging themes at the intersection of technology, innovation and policy. To get it in your inbox, subscribe here.)

EU’s AI Rules get lawmaker approval

Five years after the European Union’s AI Act regulations were proposed, EU lawmakers have given their final approval to the bloc’s AI law, and voted in favour of it. The rules could take effect this year itself, and will certainly influence how other governments worldwide choose to regulate AI technology in their own jurisdictions.

The EU AI Act takes a “risk-based approach” and is aimed at protecting consumers who are interacting with AI-powered products and services. While systems related to content recommendation or spam filters are examples of low-risk services, AI integration in the fields of healthcare and electricity will require creators to use legitimate data and be transparent. Meanwhile, some banned AI systems include those for behaviour control, types of predictive policing, and emotion recognition in certain settings. AI companies including OpenAI and Google will also have to reveal more information about the data they used to train their models, and will have to comply with EU copyright law.

OpenAI’s Sora to become available this year

OpenAI’s text-to-video generator, Sora, is likely to be released to most users within a few months, according to OpenAI’s chief technology officer, Mira Murati. Despite its limited initial release, Sora was able to deliver hyper-realistic scenes featuring people and animals, based on text prompts alone, according to OpenAI. Videos produced with the help of Sora went viral on social media platforms like X. OpenAI also plans to integrate audio and give users the ability to modify the content produced by Sora.

Murati told The Wall Street Journal that OpenAI relied on publicly available or licensed data to train Sora, and that it used content from OpenAI’s partner Shutterstock.

U.S. moves closer towards TikTok ban

The U.S. moved a step closer to potentially banning the video sharing app TikTok as the House of Representatives this week passed a bill that could result in a ban if TikTok’s Chinese parent, ByteDance, does not sell the app. A number of U.S. lawmakers and law enforcement officials have publicly declared that TikTok is a national security threat, and they believe its data can be accessed by the Chinese government to carry out espionage. The U.S. is one of TikTok’s largest markets, where it has been criticised for its addictive algorithm and data storage practices.

The bill was passed by a vote of 352-65 and will proceed to the Senate for further deliberation. More than 150 million Americans use TikTok, including small business owners. TikTok denied that it shares data with the Chinese government, even though its Chinese parent has in the past struggled with the regime’s policies and expectations.



Source link