Financial Association December 5th (Editor Niu Zhanlin)In June this 独家资讯year, the European Parliament voted the world's first "Artificial Intelligence Act". At present, the bill is facing a moment of success or failure. The negotiating representatives tried to finalize the final details this week.
According to the legislative agenda, the European Parliament will negotiate with the European Commission and member states on the bill.The bill determined by the negotiation will take effect on the 20th day from the date of announcement of the EU official communiqué, and it will be applied from 24 months after it takes effect. The more optimistic speculation is that it will be applied in 2026.
This will be the first bill in the world to comprehensively regulate artificial intelligence, which will further consolidate the EU's position as a global trend leader in controlling the technology industry.
But at the last moment, there are still controversy on how to monitor basic models such as ChatGPT, causing this process to fall into a deadlock.Large -scale technology companies are lobbying to oppose the excessive supervision of innovation, while European legislators hope to increase protection measures for the cutting -edge artificial intelligence systems that these companies are developing.
At the same time, global alliances such as the United States, the United Kingdom, and the Seven Domestic Groups (G7) also joined the competition to formulate protection measures for this technology. Researchers and human rights organizations warned that the generation of artificial intelligence constituted the survival of human beings.Danger also constitutes risks for daily life.
Political Risk Consulting Company Eurasia Group's scientific and technological policy analyst Nick Reiners said: "The EU Artificial Intelligence Act is difficult to become the global gold standard for industry supervision. It may not be able to reach an agreement before the European Parliament election next year. In the last round of negotiations on Wednesday,There are too many things to be finalized. Even if they work until late at night, they may be postponed to next year. "
When the European Commission announced the above -mentioned artificial intelligence draft in 2021, it rarely mentioned universal artificial intelligence systems such as chat robots.Instead, the artificial intelligence system is divided into four risk levels (from minimum to unacceptable), which is essentially as product safety legislation.
But with the booming development of artificial intelligence, everything has changed. It has extremely powerful functions in composition, creative images and writing articles.However, generating artificial intelligence also triggers people's concerns, that is, this technology may be used to launch large -scale network attacks or manufacture new biological weapons.
These risks have prompted EU legislators to expand their supervision content to the basic model.These systems are also called large -language models, which gives the ability to create new things such as ChatGPT and other chat robots. This is very different from traditional artificial intelligence. Traditional artificial intelligence is to process data and use predetermined rules to complete tasks.
Last month, OPENAI experienced a dramatic internal fighting incident, and then Ultraman regained OPENAI.In this regard, Titidi Bretton, director of industrial policy of the European Commission, said that at least the situation is very clear now that companies like OpenAi are defending their own business, not public interest.
At present, the power of the EU's internal opposition to the Artificial Intelligence Act is mainly France, Germany, and Italy. The three largest economies in the EU tend to adopt self -supervision methods similar to the American model.This change is considered to help French startup MISTRAL AI and Germany's Aleph Alpha and other local -generated artificial intelligence companies.
In addition, the European Union negotiating representatives still need to solve some other controversial issues, including a proposal to completely prohibit real -time public facial recognition.Some countries hope to exempt that law enforcement agencies can use it to find missing children or terrorists, but human rights organizations are worried that this will actually create a legal foundation for surveillance.
The three EU government departments will usher in one of the last opportunities for their agreement on Wednesday.Even after they reached an agreement, the European Union's 705 members still have to approve the final version.This vote needs to be carried out before April next year in order to take effect before the European Parliament elections in June next year.If they cannot complete it in time, the legislation will be postponed until later next year, and there may be new EU leadership who have different views on artificial intelligence at that time.
In a group discussion last week, Romanian parliamentarian Dragos Tudorache, who led the European Parliament's Artificial Intelligence Act, said: "It is likely that this is indeed the last good opportunity. We still need more time to negotiate. In short, the prospects are still the prospects.unclear."
版权声明：如非注明，此文章为本站原创文章，转载请注明： 转载自Express information website