AI models with systemic risks given pointers on how to comply with EU AI rules
BRUSSELS (Reuters) -The European Commission set out guidelines on Friday to help AI models it has determined have systemic risks and face tougher obligations to mitigate potential threats comply with European Union artificial intelligence regulation (AI Act).
The move aims to counter criticism from some companies about the AI Act and the regulatory burden while providing more clarity to businesses which face fines ranging from 7.5 million euros ($8.7 million) or 1.5% of turnover to 35 million euros or 7% of global turnover for violations.
The AI Act, which became law last year, will apply on Aug. 2 for AI models with systemic risks and foundation models such as those made by Google, OpenAI, Meta Platforms, Anthropic and Mistral. Companies have until August 2 next year to comply with the legislation.
The Commission defines AI models with systemic risk as those with very advanced computing capabilities that could have a significant impact on public health, safety, fundamental rights or society.
The first group of models will have to carry out model evaluations, assess and mitigate risks, conduct adversarial testing, report serious incidents to the Commission and ensure adequate cybersecurity protection against theft and misuse.
General-purpose AI (GPAI) or foundation models will be subject to transparency requirements such as drawing up technical documentation, adopt copyright policies and provide detailed summaries about the content used for algorithm training.
"With today's guidelines, the Commission supports the smooth and effective application of the AI Act," EU tech chief Henna Virkkunen said in a statement.
($1 = 0.8597 euros)
(Reporting by Foo Yun Chee;Editing by Elaine Hardcastle)
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Borneo Post
an hour ago
- Borneo Post
Sowing seeds of intelligence: Rural women fuel China's AI revolution
Wang works on AI training at Aidou Technology Co Ltd in Yijun County, Tongchuan City of northwest China's Shaanxi Province on June 12, 2025. – Xinhua photo XI'AN (July 20): Sitting in front of a computer, Wang Meimei tagged a paragraph, converted a voice message into text and marked the speaker's emotions. From the basic data processing of Deepseek to the facial recognition systems used in everyday payments, her work quietly supports the technology that powers many people's lives. But she is not, in fact, an office worker. At least not in the traditional sense. Wang, 46, from northwest China's Shaanxi Province, typically spends her days tending to fields of corn. Mondays, however, are different. At the start of every week, Wang heads to an office in Yijun County to her job as an artificial intelligence (AI) trainer. 'My job is to work as the AI's teacher, feeding it with information so that its replies are more accurate,' she said proudly. Wang's hometown, Tongchuan City, is on the Loess Plateau. In the past, her family eked out a living on the meager income generated from their 30 mu (about 2 hectares) plot of land. 'I was in a boarding school between grade 7 and 9. I would take a bag of steamed buns from home to school with me every Sunday,' she said, adding that they were her food for the entire week. 'My biggest dream then was that one day, I do not have to eat dry, moldy buns.' Like many of her fellow villagers, she left education after middle school; instead, choosing to start earning money through farming and casual work. In the following years, constant improvements to agricultural mechanisation meant she spent less and less time in the field. With her newfound free time, Wang looked for other opportunities. On February 21, 2021, she made her way to Aidou Technology Co Ltd in the county seat of Yijun, for an interview she had secured thanks to a friend's recommendation. 'I had never heard of AI before,' Wang admitted. 'During the interview, I dared not even look up.' A lack of confidence was the impression most interviewees left on Zhang Rui, the head of the company. 'When the company was in its early stage of recruitment, many of the applicants were local farmers,' she recalled. 'Some even called us to ask how to operate the elevator to reach the floor for the interview. They were nervous and many even cried when they struggled to answer our questions.' Zhang was not put off by the lack of confidence she saw in those interviewees. In fact, she saw vast untapped potential in the candidates, and felt, with a little training, they could far exceed their own expectations. Data annotation is a fast-growing industry in China, which is home to seven data annotation bases with 58,000 workers, according to the National Data Administration. Headquartered in Hangzhou in China's AI heartland of Zhejiang Province, Aidou is the first AI data annotation enterprise in Tongchuan. Zhang shared that the local government invited technology enterprises to increase employment quality, revitalise the countryside and boost the local economy. Therefore, the company kept its recruitment standards relatively flexible, considering candidates of all ages. 'What we value most is patience and the ability to learn, because this is an industry that is constantly evolving,' she said. Wang works at a corn field in Guozhai Village of Yijun County, Tongchuan City of northwest China's Shaanxi Province on June 12, 2025. – Xinhua photo With a small population of only 72,000 people, Yijun was an underdeveloped county with much of its workforce choosing to leave for work opportunities. To find suitable employees, staff from the county's employment service center visited all the townships to talk to local women, and encouraged local university students to look for opportunities in their hometown. 'Finally we found 20 plus workers,' said Bai Yanqi with the center. They also provided a venue and supporting facilities for the company free-of-charge. Today, the company has grown significantly, employing more than 240 staff members, over 70 per cent of whom are women from nearby villages. They began by learning basic computer skills and have now become certified AI trainers. The company's revenue has exceeded 35 million yuan (about US$4.9 million). 'Here for a woman like me, in her 40s, most of the job opportunities are for waitstaff or supermarket cashiers, which are shift work. 'In that case, it would be hard for me to juggle work and family,' she told Xinhua. Now earning 4,000 yuan a month with Aidou, she enjoys flexible working hours, allowing her to care for her family while also discovering a sense of value in her job. 'Before I took this job, at home I used to care for my husband, who was the main breadwinner in our family, and my children,' Wang recalled. 'Now that I have helped graduate AI from a 'primary school student' to a 'middle high school student,' my husband and children are all very supportive, and even volunteered to help out with the housework so I can focus more on my work.' She also now uses AI to help with some of her real-world problems. 'For instance, I have used Deepseek to search for corn diseases and find the right treatment,' she said. Like Wang, other rural women who used to play mahjong or square dance in their spare time have seen their lives change. The job with AI opened up a whole new world. Ma Shuaishuai, who was a nursery teacher, is now versed in AI application scenarios such as driverless technology, higher education, ecological environment, and finance and insurance. Wang Haili, once a textile worker and courier, recently bought a 100,000 yuan sedan to use for local travel. 'Their horizons have been broadened,' said Zhang Rui. 'From what I have seen, the most noticeable change for those women is that they are becoming more confident and have higher self-worth.' Some women wore slippers to the office at the beginning. Now they wear business suits. In recent years, the Chinese government has introduced a wide range of measures to accelerate AI innovation and promote its application, intending to support new industrialisation and the development of the industrial sector. China's pool of AI researchers surged from fewer than 10,000 in 2015 to over 52,000 in 2024. The country has made holistic advancements in AI development, fostering a thriving AI industrial ecosystem. It now hosts over 400 'little giant' firms – specialised small and medium-sized enterprises that excel in niche AI markets, including AI innovator DeepSeek. For Yijun County, the development of AI has created jobs for more than 1,000 people. Its potential has attracted talent to carve out a career in their hometown, like 31-year-old Zhao Yangjuan who quit her job in Xi'an, capital of Shaanxi. 'I can see more young people like me coming back, and Yijun is becoming more prosperous,' she said. 'AI is not only changing individuals like me, but also where I come from.' Wang Meimei told Xinhua that she has to work hard. 'AI is developing so fast in our country, so I have to continue learning to keep pace with the latest technology,' she said. – Xinhua AI China rural women Xian Xinhua
The Sun
10 hours ago
- The Sun
China EV brands Zeekr, Neta accused of inflating car sales
CHINESE electric vehicle brands Neta and Zeekr inflated sales in recent years to hit aggressive targets, with Neta doing so for more than 60,000 cars, according to documents reviewed by Reuters and interviews with dealers and buyers. The companies arranged for cars to be insured before they were sold to buyers, the documents show, enabling them under Chinese industry car registration practices to book sales early so they could hit the monthly and quarterly targets, the dealers and buyers said. Neta booked early sales of at least 64,719 cars through this method from January 2023 to March 2024, according to copies of records it sent to dealers, seen by Reuters. That was more than half the sales of 117,000 vehicles it reported over the 15 months. Zeekr, a premium EV brand owned by Geely, used the same method to book early sales in late 2024 in the southern city of Xiamen through its main dealer there, state-owned Xiamen C&D Automobile, according to dealers, buyers and sales receipts seen by Reuters. Vehicles booked as sold before reaching a buyer are called 'zero-mileage used cars' in the Chinese auto industry. The practice has emerged out of cutthroat competition for sales in the world's largest auto market, which is reeling from a brutal, years-long price war caused by chronic overcapacity. The industry faces a moment of reckoning, with state media calling out the zero-mileage car practice, the cabinet pledging to regulate 'irrational' competition, and other central government bodies organising meetings with the industry's largest players to express concern about such methods. On Saturday a publication run by the China Association of Auto Manufacturers said the industry ministry was planning to clamp down on the practice by banning cars from being resold within six months of being registered as a sale. STATE MEDIA FOCUS Also on Saturday, state media reported that Zeekr had been selling cars with insurance already purchased to inflate sales, the first such naming and shaming of a specific automaker. In a front-page story, the China Securities Journal newspaper interviewed Zeekr car buyers in cities such as Guangzhou and Chongqing, who the newspaper said had found that their cars already had insurance policies before they were sold. They said they were refused refunds, even though they felt they were deceived. The newspaper questioned Zeekr's unusually high sales in the cities of Shenzhen and Xiamen in December. Its reported sales in Xiamen surged to 2,737 that month, more than 14 times its monthly average. Reuters could not determine how much of that volume might have been booked early. The China Securities Journal also raised questions over Neta's sales, saying it showed anomalies. Reuters is reporting for the first time details of how Neta inflated sales. Zeekr, Zhejiang Hozon New Energy Automobile, which owns Neta, and Xiamen C&D did not respond to requests for comment on Saturday. A spokesperson for Geely said, 'Geely firmly rejects the report put forward by the China Securities Journal.' The spokesperson declined to comment on Reuters findings or provide further details. Li Yanwei, an analyst with the China Automobile Dealers Association, said he believed the firms carried out such practices to embellish their financial reports and achieve their performance goals. 'This way of whitewashing performance is not advisable,' he wrote on Chinese social media platform Weibo on Saturday. Analysts and investors tracking China's auto industry gauge performance and estimate inventory levels with two sets of sales data. Wholesale numbers reported by automakers to the industry association show sales from automakers to dealers, while retail data compiled from insurance registration records show the sales to users. Some zero-mileage used cars are exported to be sold as second-hand cars overseas, but analysts and dealers say the domestic sales volume is significantly higher, with Chinese customers nationwide buying what they believe to be discounted new vehicles, only to find out later their car is not insured under their name. PRESSURE ON DEALERS Last month the state-owned People's Daily, the mouthpiece of China's ruling Communist Party, published an editorial condemning the sale of zero-mileage used cars domestically and listing a litany of harms the practice brings upon the industry and buyers. This month four dealer associations based in the wealthy Yangtze River Delta urged automakers to set them more reasonable sales targets and incentive policies, saying, without providing details, that dealers were being forced to falsify sales. Neta booked sales early by arranging insurance policies for cars before sending them to dealers, according to records shared with Reuters and a dealer for the brand. The records contain details for each car and the insurance policies purchased on them, with the names of the insurance agents. Dealers were able to refer to these when they found a buyer to transfer the policy to, according to copies seen by Reuters. The company booked early sales of 64,719 cars this way. 'In Neta's case, the company made it clear to dealers that the cars were insured ahead of time and therefore counted as sold,' said the dealer, who spoke on condition of anonymity, citing fears of retaliation from the company. 'We had to explain to buyers that the traffic insurance was complementary and remind them it would expire earlier and should be renewed on time,' he said. But three Neta buyers, who asked not to be named, told Reuters the dealerships had not told them the policies had begun well before the purchase date, only finding out when the policies expired. The dealer said Neta started doing this in late 2022 to obtain EV subsidies that were set to end that year. Neta's sales peaked in 2022 when it was ranked as the eighth-largest maker of new EVs in China with sales of 152,000 vehicles. Sales fell last year to 87,948 vehicles, including 23,399 exported, and it sold 1,215 cars in the first quarter of 2025, according to data from the China Association of Automobile Manufacturers. The brand has been in financial trouble since late 2024, and its owner, Zhejiang Hozon New Energy Automobile, entered bankruptcy proceedings in China last month, according to state media. 'JUST DO IT' The Neta dealer said many of the zero-mileage used cars he received from the company remained in his warehouse, unsold. The company 'only had one message: Just do it, everyone else is doing it'. Zeekr, which is being privatised by Geely Auto, booked sales with the help of Xiamen C&D, which runs dealerships for Zeekr and other brands. Xiamen C&D insured and registered the vehicles under the names of two subsidiaries in December, allowing Zeekr to count the sales before year-end, according to four dealers and two buyers, as well as a receipt shared with Reuters. Zeekr dealers sold some of the cars in subsequent months to buyers in other cities such as Beijing and Chongqing, the sources said. 'The Zeekr salesman said the car would be 3,000 yuan ($420) less than a car I would get from the store and I would also get a charging coupon worth 10,000 yuan,' said a buyer in another southern city. He declined to be named, citing concerns of retaliation from the automaker. The China Securities Journal reported that most of the owners it spoke to said their cars were insured by Xiamen C&D and its affiliates. China Automobile Dealers Association data showed that 2,508 of the 2,737 sales Zeekr booked in Xiamen in December were sold to companies, while 257 went to individual buyers. But data published by Xiamen's vehicle administration bureau showed just 271 cars registered in December for license plates, which genuine buyers generally obtain once they receive their cars. - Reuters
The Sun
15 hours ago
- The Sun
Greek olive oil producers pivot from US amid Trump tariff threats
VARVASAINA: Greek olive oil producer Konstantinos Papadopoulos acted swiftly when U.S. President Donald Trump hinted at imposing tariffs on European goods earlier this year. His family-run business quickly secured a new buyer in Brazil, a market traditionally dominated by Portuguese olive oil. A shipment of 15,000 bottles is set to arrive in Itapoa soon, with another deal in Australia nearing completion. 'I think we learned a lesson from Trump not to rely with all our strength on one market... and to always have alternatives,' Papadopoulos said during a visit to his mill, where tanks brimmed with olive oil ready for export. The proposed 30% tariff on EU products has rattled industries from wine to automobiles, pushing producers like Papadopoulos to explore new markets preemptively. Greece, the fifth-largest olive oil exporter to the U.S., ships 8,000-10,000 tonnes annually. Other major European producers—Spain, Italy, and Portugal—face similar challenges. For Greece, olive oil is a cornerstone of agriculture, with ancient groves dotting its landscapes. The Papadopoulos family exported 350 tonnes to the U.S. in 2024, a third of their total output. This year, shipments have dropped to 100 tonnes, with a projected 40% decline if tariffs take effect. 'While we had built all our infrastructure based on a large scale on the American market, suddenly we see it disappearing,' Papadopoulos said. - REUTERS
