EU lawmakers have agreed a tough set of rules over the use of AI, including restrictions on chatbots such as ChatGPT, as Brussels pushes forward in enacting the world’s most restrictive regime on the development of the technology.

歐盟立法者已經(jīng)就人工智能使用制定了一系列嚴(yán)格的規(guī)則，包括對(duì)ChatGPT等聊天機(jī)器人的限制。布魯塞爾正在推動(dòng)實(shí)施全球最嚴(yán)格的技術(shù)開(kāi)發(fā)監(jiān)管體系。

MEPs on Thursday voted in favour of a sweeping set of regulations that will form part of Europe’s Artificial Intelligence Act, including a total ban of almost all facial recognition for surveillance of citizens.

周四，歐洲議會(huì)投票通過(guò)了一項(xiàng)廣泛的法規(guī)，這將成為歐洲《人工智能法案》的一部分，從總體上幾乎禁止了所有用于公民監(jiān)控的面部識(shí)別。

Under the European parliament’s proposals, developers of generative AI models like ChatGPT would have to disclose content that was generated by AI and publish summaries of copyrighted data used for training purposes, so that creators can be remunerated for the use of their work.

根據(jù)歐洲議會(huì)提出的建議，像ChatGPT這樣生成式AI模型開(kāi)發(fā)者必須披露由AI生成的內(nèi)容，并發(fā)布用于培訓(xùn)目的版權(quán)數(shù)據(jù)摘要，以便創(chuàng)作者可以獲得他們作品使用費(fèi)。

The agreement between MEPs, who have been fighting over measures to police artificial intelligence for close to two years, will kick-start negotiations between EU member states, the European Commission and the European parliament in the year ahead.

近兩年來(lái)歐洲議會(huì)議員一直在就人工智能監(jiān)管措施展開(kāi)爭(zhēng)論，此次協(xié)議將在未來(lái)一年內(nèi)啟動(dòng)歐盟成員國(guó)、歐洲委員會(huì)和歐洲議會(huì)之間關(guān)于該問(wèn)題的談判。

The proposed law comes amid growing concerns around the world about the potential abuses of the fast-developing technology.
Last week, the US Federal Trade Commission fired a warning shot to the industry, saying it was “focusing intensely on how companies may choose to use AI technology, including new generative AI tools, in ways that can have actual and substantial impact on consumers”. The UK competition watchdog is also launching a review of the AI market.

在擬議中的法律出臺(tái)之際，世界各地越來(lái)越擔(dān)心快速發(fā)展的技術(shù)可能被濫用。上周，美國(guó)聯(lián)邦貿(mào)易委員會(huì)向該行業(yè)發(fā)出警告，稱其“正在密切關(guān)注公司如何選擇使用人工智能技術(shù)，包括新的生成人工智能工具，以對(duì)消費(fèi)者產(chǎn)生實(shí)際和實(shí)質(zhì)性影響的方式” 。英國(guó)競(jìng)爭(zhēng)監(jiān)管機(jī)構(gòu)也正在對(duì)人工智能市場(chǎng)進(jìn)行審查。

Drago? Tudorache, an MEP who led the development of the proposals, told journalists on Wednesday that the agreed text had been among the most negotiated in the parliament’s history, with more than 1,000 amendments.

領(lǐng)導(dǎo)該提案制定的歐洲議會(huì)議員Drago? Tudorache周三告訴記者，此次商定的文本是議會(huì)歷史上協(xié)商最多的文本之一，有 1,000 多項(xiàng)修正案。

“[The AI Act] is the first legislation of this kind worldwide,” said Tudorache. “The EU can lead the way, globally, in making AI human-centric, trustworthy and safe.”

Tudorache說(shuō)：“《人工智能法案》是全球首部此類立法。歐盟可以在全球范圍內(nèi)帶頭使人工智能以人為本、值得信賴和安全。”

Member states and legislators are likely to enter drawn-out discussions on whether live facial recognition is acceptable in limited circumstances such as when there is a terrorist threat, people with knowledge of the discussions said.

知情人士表示，歐盟成員國(guó)和立法者可能會(huì)就在有限情況下（例如存在恐怖威脅時(shí)）是否可以接受實(shí)時(shí)面部識(shí)別展開(kāi)曠日持久的討論。

Restrictions on real-time biometrics is an area that will demand “tough” negotiations and on which the EU will probably “need to find a solution”, said MEP Brando Benifei, who developed the proposals with Tudorache.

與 Tudorache 一起制定提案的歐洲議會(huì)議員 Brando Benifei 表示，對(duì)實(shí)時(shí)生物識(shí)別技術(shù)的限制是一個(gè)需要“艱難”談判的領(lǐng)域，歐盟可能“需要找到解決方案”。
原創(chuàng)翻譯：龍騰網(wǎng) http://nxnpts.cn 轉(zhuǎn)載請(qǐng)注明出處


“Governments are speaking clearly about their willingness to use [these systems],” he added.

他補(bǔ)充道：“政府明確表態(tài)愿意使用這些（識(shí)別）系統(tǒng)?！?/b>

The parliament has agreed to retrospective “biometric identification systems” for the prosecution of serious crimes, if police forces gain the approval of a judge.

在法官批準(zhǔn)的情況下，議會(huì)已同意警察使用回顧性“生物特征識(shí)別系統(tǒng)”以起訴嚴(yán)重犯罪。

But Benifei said he was optimistic that member states would be supportive of the parliament’s proposals to legislate on generative AI models.

但Benifei表示他樂(lè)觀地認(rèn)為，成員國(guó)將支持議會(huì)就生成式AI模型立法的提案。
原創(chuàng)翻譯：龍騰網(wǎng) http://nxnpts.cn 轉(zhuǎn)載請(qǐng)注明出處


The European parliament is expected to endorse the rules in June, before negotiations on the final form of the law begin with other EU institutions.

預(yù)計(jì)歐洲議會(huì)將在6月份批準(zhǔn)這些規(guī)則，然后與其他歐盟機(jī)構(gòu)開(kāi)始就法律的最終形式進(jìn)行談判。