European Union lawmakers gave final approval to the 27-nation bloc’s artificial intelligence law Wednesday last week, putting the world-leading rules on track to take effect later this year.
Lawmakers in the European Parliament voted overwhelmingly in favor of the Artificial Intelligence Act, five years after regulations were first proposed. The AI Act is expected to act as a global signpost for other governments grappling with how to regulate the fast-developing technology.
Here’s a look at the world’s first comprehensive set of AI rules:
Photo: AP 照片：美聯社
HOW DOES THE AI ACT WORK?
Like many EU regulations, the AI Act was initially intended to act as consumer safety legislation, taking a “risk-based approach” to products or services that use artificial intelligence.
The riskier an AI application, the more scrutiny it faces. The vast majority of AI systems are expected to be low risk, such as content recommendation systems or spam filters. Companies can choose to follow voluntary requirements and codes of conduct.
Photo: AP 照片：美聯社
High-risk uses of AI, such as in medical devices or critical infrastructure like water or electrical networks, face tougher requirements like using high-quality data and providing clear information to users.
Some AI uses are banned because they’re deemed to pose an unacceptable risk, like social scoring systems that govern how people behave, some types of predictive policing and emotion recognition systems in school and workplaces.
Other banned uses include police scanning faces in public using AI-powered remote “biometric identification” systems, except for serious crimes like kidnapping or terrorism.
Photo: Reuters 照片：路透
WHAT ABOUT GENERATIVE AI?
The law’s early drafts focused on AI systems carrying out narrowly limited tasks, like scanning resumes and job applications. The astonishing rise of general purpose AI models, exemplified by OpenAI’s ChatGPT, sent EU policymakers scrambling to keep up.
They added provisions for so-called generative AI models, the technology underpinning AI chatbot systems that can produce unique and seemingly lifelike responses, images and more.
Developers of general purpose AI models — from European startups to OpenAI and Google — will have to provide a detailed summary of the text, pictures, video and other data on the Internet that is used to train the systems as well as follow EU copyright law.
AI-generated deepfake pictures, video or audio of existing people, places or events must be labeled as artificially manipulated.
There’s extra scrutiny for the biggest and most powerful AI models that pose “systemic risks,” which include OpenAI’s GPT4 — its most advanced system — and Google’s Gemini.
The EU says it’s worried that these powerful AI systems could “cause serious accidents or be misused for far-reaching cyberattacks.” They also fear generative AI could spread “harmful biases” across many applications, affecting many people.
Companies that provide these systems will have to assess and mitigate the risks; report any serious incidents, such as malfunctions that cause someone’s death or serious harm to health or property; put cybersecurity measures in place; and disclose how much energy their models use.
Brussels first suggested AI regulations in 2019, taking a familiar global role in ratcheting up scrutiny of emerging industries, while other governments scramble to keep up.
The AI Act is expected to officially become law by May or June, after a few final formalities, including a blessing from EU member countries.
Violations of the AI Act could draw fines of up to 35 million euros (US$38 million), or 7 percent of a company’s global revenue.
(AP)
上週三，歐盟人工智慧法獲歐盟27個會員國的立法者最終批准，這領先全球的法規預計在今年稍晚生效。
該法案首次提出五年後，歐洲議會議員以壓倒性多數投票通過《人工智慧法》。人工智慧科技日新月異，各國政府無不想方設法要加以監管，《人工智慧法》預計將為全球指出一條道路。
這世界第一部全面的人工智慧法規，其梗概如下：
《人工智慧法》如何運作？
跟許多歐盟法規一樣，《人工智慧法》的立法初衷是保障消費者安全，對使用人工智慧的產品或服務採取「依風險級別管理之方法」。
人工智慧應用程式的風險越大，所受的審查也就越多。絕大多數人工智慧系統預計都是低風險的，例如內容推薦系統或垃圾郵件過濾器。公司可以選擇自發遵循法規及行為準則。
人工智慧的高風險用途，例如在醫療設備、供水或電網等關鍵基礎設施中，所面對的規定更嚴格，例如必須使用高品質數據，以及向用戶提供清楚的資訊。
某些人工智慧的用途則被禁止，因被認為會帶來不可接受的風險，例如控制人們行為的社會評分系統、學校和工作場所某些類型的預測性警務與情緒辨識系統。
其他被禁止的用途還包括警察使用人工智慧遠端「生物辨識」系統在公共場合掃描臉部，但偵查綁架或恐怖攻擊等嚴重犯罪則不在此限。
生成式人工智慧如何規範？
這部法律的早期草案著重於人工智慧系統執行有限的任務，例如掃描履歷和求職文件。以OpenAI的ChatGPT為代表的通用人工智慧模型的崛起勢如破竹，讓歐盟政策制定者爭先恐後地跟上。
因此他們增加了所謂生成式人工智慧模型的規定，該模型是支撐人工智慧聊天機器人系統的技術，可產生獨特且看似真人的回應、圖像等。
通用人工智慧模型的開發者——從歐洲新創公司到OpenAI和谷歌——必須在網路上提供用於訓練系統的文字、圖片、影片和其他資料的詳細摘要，並遵守歐盟版權法。
以人工智慧生成的現世人物、地點或事件的深度偽造圖片、影片或音訊，必須標記為經人為變造。
「系統性風險」最廣、最強大的人工智慧模型，則須進行額外的審查，包括OpenAI的GPT4（其最先進的系統）和谷歌的Gemini。
歐盟表示擔心這些強大的人工智慧系統可能「導致嚴重事故或被濫用於影響深遠的網路攻擊」。他們還擔心生成式人工智慧可能會在許多應用程式中傳播「有害偏見」，從而影響許多人。
提供這些系統的公司必須評估與減輕風險、報告所有的嚴重事件，例如導致死亡或嚴重損害健康或財產的故障，制定網路安全措施，並披露其模型的能源使用量。
布魯塞爾於2019年首次提出人工智慧法規，如同歐盟在全球常扮演的對新興產業加強審查的領頭羊角色，他國政府則紛紛跟上。
經過一些最終程序（包括歐盟成員國的批准）後，《人工智慧法》預計將在5月或6月正式成為法律。
違反《人工智慧法》可能會被罰款最高3,500萬歐元（3,800萬美元），或該公司全球收入的7%。
（台北時報林俐凱編譯）
A: There are always adventurers who want to conquer Taipei 101 as a world-class landmark. Didn’t someone once parachute from the top of it? B: Yeah, that’s right. Austrian extreme sportsman Felix Baumgartner once parachuted from the rooftop observation deck in 2007 without permission. He died earlier last year in a powered paragliding crash at the age of 56. A: Hollywood superstar Tom Cruise also almost jumped off Taipei 101 for “Mission Impossible 3.” B: What? But I didn’t see the building in the movie. A: The news says that the film’s producers applied to the Taipei City
People use far more than just spoken language to communicate. Apart from using our voices to pronounce words, we also use body language, which includes countless facial expressions. Most people know that smiles and frowns indicate pleasure and displeasure, or that wide eyes with raised eyebrows typically show surprise. However, there is a lot more to learn about how facial expressions can help or hinder communication. People often unintentionally reveal their emotions through very tiny facial movements known as “microexpressions.” The term was popularized by psychologist Paul Ekman, who found that people from cultures across the world generally recognize
Nestled in the mountain region of northern Taiwan lies a small village that was once bustling due to a thriving coal industry, only to later gain fame for an entirely different reason — cats. This is Houtong Cat Village (猴硐貓村), located in Ruifang District, New Taipei City. Traditional Coal Mining and Village Transformation Houtong was originally known as “Monkey Cave” (houtong, 猴洞), a name derived from the wild monkeys that once inhabited caves in the surrounding hills. During the Japanese colonial period in the early 20th century, rich coal deposits were discovered and developed, rapidly turning Houtong into one of northern Taiwan’s
對話 Dialogue 清清：氣象預報說週末好像有寒流，天氣這麼冷，我只想宅在家裡。 Qīngqīng: Qìxiàng yùbào shuō zhōumò hǎoxiàng yǒu hánliú, tiānqì zhème lěng, wǒ zhǐ xiǎng zhái zài jiālǐ. 華華：說到這個，我週末想約剛認識的新朋友出去玩，但寒流來的話戶外活動可能不太合適，我還不知道要做什麼好。 Huáhua: Shuō dào zhège, wǒ zhōumò xiǎng yuē gāng rènshí de xīn péngyǒu chūqù wán, dàn hánliú lái de huà hùwài huódòng kěnéng bú tài héshì, wǒ hái bù zhīdào yào zuò shénme hǎo. 清清：要不要考慮去玩密室脫逃？我以前跟朋友去過，很好玩，而且不怕風吹雨淋。 Qīngqīng: Yào bú yào kǎolǜ qù wán mìshì tuōtáo? Wǒ yǐqián gēn péngyǒu qù guò, hěn hǎowán, érqiě bú pà fēng chuī yǔ lín. 華華：我覺得不錯，可是我們才剛認識，會不會太尷尬？ Huáhua: Wǒ juéde búcuò, kěshì wǒmen cái gāng rènshí, huì bú huì tài gāngà? 清清：不會啦！玩密室脫逃需要大家一起合作，反而更容易變熟。 Qīngqīng: Bú huì la! Wán mìshì tuōtáo xūyào dàjiā yìqǐ hézuò, fǎn’ér gèng róngyì
RSS