Why read this: This piece gives Year 11 to 13 readers a high-stakes example of critical journalism in action. Students who can already follow surface meaning at C1 will be pushed to evaluate an argument that itself interrogates corporate truth claims, with the AI industry as a vivid and current case.
What to notice: The argument is carried by sustained irony, a multi-voice cast of expert critics (Vallor, Bender, Khlaaf), and a mythic framing (salvation, apocalypse, gods) layered over corporate critique. Watch for hedged modals (could, might, may), reported speech, and ironic understatement, which together signal where the writer stands without saying so directly.
Skills practised: Readers practise tracking attribution chains across five named voices, recognising rhetorical framing in journalistic prose, and weighing evidence quality between company spokespeople and independent experts. The closing parallels to earlier Silicon Valley hype cycles train students to evaluate analogy as argument.
The fear they sell: how AI firms market the apocalypse
Anthropic, OpenAI and the strange commercial logic of warning the world about your own product.
Tap any green word in the article to see its meaning.
Notice a pattern in the technology press: a company unveilsunveil: to reveal or make known something previously secret or hidden — 揭幕；公布；揭露 an AI so formidableformidable: Causing fear, dread, awe, or discouragement as a result of size, strength, or some other impressive feature; commanding respect; causing wonder or astonishment — 因体量、力量或其他令人印象深刻的特征而令人畏惧的；令人敬畏的 that, by its own admissionadmission: The act of being allowed to enter a place or institution, or a statement acknowledging a fact. — 录取；入场；承认, it could prove civilisationcivilisation: Human society, particularly civil society — 人类社会；文明世界-ending in the wrong hands: in the possession of people who would use it for harm — 落入恶人手中. Anthropic, the firm behind Claude Mythos, struck this note in an April blog post, warning of world-altering consequences and a falloutfallout: The event of small airborne particles falling to the ground in significant quantities as a result of major industrial activity, volcano eruption, sandstorm, nuclear explosion, etc — 放射性沉降物 for economies, public safety and national security that "could be severe". Breathless commentatorscommentator: A person who comments on events, especially on television, radio, or in newspapers; an analyst or critic. — 评论员；解说员 predicted families would soon need to swap out every WiFi-enabled applianceappliance: The act of applying — 应用；运用 in the kitchen. Anthropic has, of course, decided to sell Mythos anyway.
Researchers have raised doubts about the technical claims, but let us set that aside: to ignore or postpone discussing something for the moment — 暂且不谈. The deeper oddityoddity: An odd or strange thing or opinion — 奇特之处；怪事 is the genre. McDonald's does not advertise a burger so delicious that grilling it would be unethicalunethical: Not morally approvable; morally bad; not ethical — 不道德的；违反职业道德的, yet executives at the leading AI providers issue exactly this kind of warning monthly. Critics argue the strategy works because it keeps audiences fixatedfixate: to attend to something to the exclusion of all others; used with on — 执着于；念念不忘 on apocalypse and distracts from harms already underway. What looks like candourcandour: The state of being sincere and open in speech; honesty in expression — 坦率；直言不讳 is fear mongering: deliberately spreading fear, often to influence people for one's own advantage — 贩卖恐惧 that inflates the technology, lifts share prices, and pushes a narrative in which regulatorsregulator: A person or group that sets standards of practice, especially those established by law — 监管机构 must stand aside: to step back and not intervene — 袖手旁观 / 不插手 because only the firms can be trusted to govern what they have built. "If you portrayportray: to depict or represent someone or something in a particular way in art, literature, or media — 描绘；刻画；扮演 these technologies as somehow almost supernaturalsupernatural: Not of the usual; not natural; altered by forces that are not understood fully if at all — 异常的；非寻常的 in their danger, it makes us feel like we are powerlesspowerless: Lacking sufficient power or strength — 无力的；无能为力的, like we are outmatchedoutmatch: to surpass or be better than something or someone else — 胜过；强过," says Shannon Vallor, professor of the ethics of AI at the University of Edinburgh.
The playbookplaybook: (by extension) A set of commonly employed tactics and strategies — 策略指南 is older than Mythos. In 2019, Dario Amodei, then an OpenAI executive and now Anthropic's chief, helped declare GPT-2 too risky for release; the company shipped it months later. Sam Altman, who criticisedcriticise: To find fault (with something) — 批评 Anthropic's fear-based marketing: promoting a product by emphasising danger or risk to the audience — 恐惧式营销 on a recent podcast, told an audience in 2015 that AI "will probably most likely lead to the end of the world, but in the meantimemeantime: the intervening period of time between two events — 其间；与此同时, there'll be great companies." In 2023, leaders including Altman, Amodei, Bill Gates and Demis Hassabis of Google DeepMind placedplace: to put (an object or person) in a specific location — 放置；置于（某位置） AI extinctionextinction: The action of making or becoming extinct; annihilation — 灭绝；消亡 alongside pandemics and nuclear war; the same year, mogulsmogul: A rich or powerful person; a magnate — 大亨；巨头 such as Elon Musk demanded a six-month pause, and Musk launched xAI inside that window. Anthropic itself began as a splinter group: a small group that has broken away from a larger organisation due to disagreement — 分裂出去的小派系 of safety-mindedminded: Having a mind (inclination) for something or a certain way of thinking about things — 有……倾向的；有……心思的 OpenAI staff, founded as a non-profit; both firms are now racing to become publicly traded. For Emily M Bender, a computationalcomputational: Of or relating to computation — 计算的；运算的 linguisticslinguistics: The scientific study of language — 语言学 professor at the University of Washington and co-author of The AI Con, the choreographychoreography: The art of creating, arranging and recording the dance movements of a work, such as a ballet — 创作、编排和记录舞蹈动作的艺术，如芭蕾舞的编排 is a pattern of unsubstantiatedunsubstantiated: Lacking substantiation; without evidence — 未经证实的；无证据支持的 claims of power. The companies, she says, are telling the world look over here: to direct attention to one thing in order to distract from another — 看这边（声东击西）, never mind the labour exploitationexploitation: The unfair treatment or use of someone or something for one's own advantage; utilisation of resources — 剥削；利用；开发 or the systems they are quietly reshaping.
The technical claims invite the same scrutinyscrutiny: careful and thorough examination or investigation of something — 仔细审查；详细检查. Anthropic reports Mythos has surfaced thousands of "high-severityseverity: The degree of something undesirable; badness or seriousness — 严重程度；严重性" vulnerabilitiesvulnerability: the quality of being exposed to the possibility of harm or damage — 脆弱性；易受攻击性；弱点 and partnered with 40-odd organisations to patchpatch: To make the changes a patch describes; to apply a patch to the files in question. Hence: — （计算机）打补丁；为……应用补丁 foundationalfoundational: Fundamental or underlying — 根本的；起根基作用的 systems that, a spokesperson said, represent the lion's share: the largest part of something — 最大的份额 of the problem. Heidy Khlaaf, chief AI scientist at the AI Now Institute, who has spent her career auditingauditing: The act or result of performing an audit — 审计；审查 exactly the kind of code analysis tools Anthropic claims to have outclassedoutclass: to surpass something or somebody else, so as to appear to be in a higher class — 远胜于；超越, is unmoved. Her biggest red flag was the absenceabsence: The state of being away from a place or the lack of something. — 缺席；缺乏；不在 of a false positive rate: the proportion of times a tool wrongly identifies something as a problem when it isn't — 误报率, the industry-standard metricmetric: A measure for something; a means of deriving a quantitative measurement or approximation for an otherwise qualitative phenomenon — 衡量指标；度量 for how often a tool flags an innocent file as a threat. Anthropic did not publish one and sidesteppedsidestep: to avoid a difficult question, problem, or person by acting indirectly — 回避 the question when she pressedpress: to exert weight or force against, to act upon with force or weight; to exert pressure upon — 施压；强行追问 for comment. "Mythos might be capable," Khlaaf says. What she rejects is the surrounding mythologymythology: A similar body of myths concerning an event, person or institution — 围绕某事物的神话故事；传说: "there are a lot of cracks in this narrative that Mythos is all powerful, we can't release it."
"To understand how a corporation will behave, look at what its incentives are," Vallor observes. The incentives are visible. Google has dropped its red lines: limits or rules a person or organisation has promised not to cross — 底线 / 红线 around AI weapons. OpenAI has fought a legal battle to shedshed: to get rid of something unwanted; to cause blood, tears, or light to flow or fall — 去除；流（泪）；脱落；散发（光） non-profit status. Anthropic has abandoned its flagshipflagship: The most important one out of a related group. — 旗舰（最重要的一家／一款） policy of never training a model it could not guarantee was safe. Gas-powered datacenterdatacenter: A facility used to house computer systems and associated components, such as telecommunications and storage equipment — 数据中心 campuses are projected to emitemit: to send out or give off — 排放；散发 more greenhouse gas than entire countries, and deepfakesdeepfake: A photo or video created using artificial intelligence, that imposes the face of one person onto the body of another — 深度伪造（影像） have crossed the point of no return: a stage past which it is impossible to undo or reverse a situation — 无法回头的临界点. In almost the same breath as their warnings of apocalypse, the executives promise salvationsalvation: The process of being saved, the state of having been saved (from hell) — 拯救；救赎: a utopia-gifting country of geniuses inside a datacenter, the climate fixed, a colony on another world. Utopia and apocalypse, Vallor argues, are two faces of one mythicmythic: Larger-than-life — 宏大的；传奇般的 coin, pitched on a scale too grand for courts or regulators to feel they have any purchase. "Nothing about them is ungovernableungovernable: Not governable — 无法管理的；不受控制的," she says. "Unless we choose not to govern them."
Notice a pattern in the technology press: a company unveilsunveil: to reveal or make known something previously secret or hidden — 揭幕；公布；揭露 an AI so formidableformidable: Causing fear, dread, awe, or discouragement as a result of size, strength, or some other impressive feature; commanding respect; causing wonder or astonishment — 因体量、力量或其他令人印象深刻的特征而令人畏惧的；令人敬畏的 that, by its own admissionadmission: The act of being allowed to enter a place or institution, or a statement acknowledging a fact. — 录取；入场；承认, it could prove civilisationcivilisation: Human society, particularly civil society — 人类社会；文明世界-ending in the wrong hands: in the possession of people who would use it for harm — 落入恶人手中. Anthropic, the firm behind Claude Mythos, struck this note in an April blog post, warning of world-altering consequences and a falloutfallout: The event of small airborne particles falling to the ground in significant quantities as a result of major industrial activity, volcano eruption, sandstorm, nuclear explosion, etc — 放射性沉降物 for economies, public safety and national security that "could be severe". Breathless commentatorscommentator: A person who comments on events, especially on television, radio, or in newspapers; an analyst or critic. — 评论员；解说员 predicted families would soon need to swap out every WiFi-enabled applianceappliance: The act of applying — 应用；运用 in the kitchen. Anthropic has, of course, decided to sell Mythos anyway.
Researchers have raised doubts about the technical claims, but let us set that aside: to ignore or postpone discussing something for the moment — 暂且不谈. The deeper oddityoddity: An odd or strange thing or opinion — 奇特之处；怪事 is the genre. McDonald's does not advertise a burger so delicious that grilling it would be unethicalunethical: Not morally approvable; morally bad; not ethical — 不道德的；违反职业道德的, yet executives at the leading AI providers issue exactly this kind of warning monthly. Critics argue the strategy works because it keeps audiences fixatedfixate: to attend to something to the exclusion of all others; used with on — 执着于；念念不忘 on apocalypse and distracts from harms already underway. What looks like candourcandour: The state of being sincere and open in speech; honesty in expression — 坦率；直言不讳 is fear mongering: deliberately spreading fear, often to influence people for one's own advantage — 贩卖恐惧 that inflates the technology, lifts share prices, and pushes a narrative in which regulatorsregulator: A person or group that sets standards of practice, especially those established by law — 监管机构 must stand aside: to step back and not intervene — 袖手旁观 / 不插手 because only the firms can be trusted to govern what they have built. "If you portrayportray: to depict or represent someone or something in a particular way in art, literature, or media — 描绘；刻画；扮演 these technologies as somehow almost supernaturalsupernatural: Not of the usual; not natural; altered by forces that are not understood fully if at all — 异常的；非寻常的 in their danger, it makes us feel like we are powerlesspowerless: Lacking sufficient power or strength — 无力的；无能为力的, like we are outmatchedoutmatch: to surpass or be better than something or someone else — 胜过；强过," says Shannon Vallor, professor of the ethics of AI at the University of Edinburgh.
The playbookplaybook: (by extension) A set of commonly employed tactics and strategies — 策略指南 is older than Mythos. In 2019, Dario Amodei, then an OpenAI executive and now Anthropic's chief, helped declare GPT-2 too risky for release; the company shipped it months later. Sam Altman, who criticisedcriticise: To find fault (with something) — 批评 Anthropic's fear-based marketing: promoting a product by emphasising danger or risk to the audience — 恐惧式营销 on a recent podcast, told an audience in 2015 that AI "will probably most likely lead to the end of the world, but in the meantimemeantime: the intervening period of time between two events — 其间；与此同时, there'll be great companies." In 2023, leaders including Altman, Amodei, Bill Gates and Demis Hassabis of Google DeepMind placedplace: to put (an object or person) in a specific location — 放置；置于（某位置） AI extinctionextinction: The action of making or becoming extinct; annihilation — 灭绝；消亡 alongside pandemics and nuclear war; the same year, mogulsmogul: A rich or powerful person; a magnate — 大亨；巨头 such as Elon Musk demanded a six-month pause, and Musk launched xAI inside that window. Anthropic itself began as a splinter group: a small group that has broken away from a larger organisation due to disagreement — 分裂出去的小派系 of safety-mindedminded: Having a mind (inclination) for something or a certain way of thinking about things — 有……倾向的；有……心思的 OpenAI staff, founded as a non-profit; both firms are now racing to become publicly traded. For Emily M Bender, a computationalcomputational: Of or relating to computation — 计算的；运算的 linguisticslinguistics: The scientific study of language — 语言学 professor at the University of Washington and co-author of The AI Con, the choreographychoreography: The art of creating, arranging and recording the dance movements of a work, such as a ballet — 创作、编排和记录舞蹈动作的艺术，如芭蕾舞的编排 is a pattern of unsubstantiatedunsubstantiated: Lacking substantiation; without evidence — 未经证实的；无证据支持的 claims of power. The companies, she says, are telling the world look over here: to direct attention to one thing in order to distract from another — 看这边（声东击西）, never mind the labour exploitationexploitation: The unfair treatment or use of someone or something for one's own advantage; utilisation of resources — 剥削；利用；开发 or the systems they are quietly reshaping.
The technical claims invite the same scrutinyscrutiny: careful and thorough examination or investigation of something — 仔细审查；详细检查. Anthropic reports Mythos has surfaced thousands of "high-severityseverity: The degree of something undesirable; badness or seriousness — 严重程度；严重性" vulnerabilitiesvulnerability: the quality of being exposed to the possibility of harm or damage — 脆弱性；易受攻击性；弱点 and partnered with 40-odd organisations to patchpatch: To make the changes a patch describes; to apply a patch to the files in question. Hence: — （计算机）打补丁；为……应用补丁 foundationalfoundational: Fundamental or underlying — 根本的；起根基作用的 systems that, a spokesperson said, represent the lion's share: the largest part of something — 最大的份额 of the problem. Heidy Khlaaf, chief AI scientist at the AI Now Institute, who has spent her career auditingauditing: The act or result of performing an audit — 审计；审查 exactly the kind of code analysis tools Anthropic claims to have outclassedoutclass: to surpass something or somebody else, so as to appear to be in a higher class — 远胜于；超越, is unmoved. Her biggest red flag was the absenceabsence: The state of being away from a place or the lack of something. — 缺席；缺乏；不在 of a false positive rate: the proportion of times a tool wrongly identifies something as a problem when it isn't — 误报率, the industry-standard metricmetric: A measure for something; a means of deriving a quantitative measurement or approximation for an otherwise qualitative phenomenon — 衡量指标；度量 for how often a tool flags an innocent file as a threat. Anthropic did not publish one and sidesteppedsidestep: to avoid a difficult question, problem, or person by acting indirectly — 回避 the question when she pressedpress: to exert weight or force against, to act upon with force or weight; to exert pressure upon — 施压；强行追问 for comment. "Mythos might be capable," Khlaaf says. What she rejects is the surrounding mythologymythology: A similar body of myths concerning an event, person or institution — 围绕某事物的神话故事；传说: "there are a lot of cracks in this narrative that Mythos is all powerful, we can't release it."
"To understand how a corporation will behave, look at what its incentives are," Vallor observes. The incentives are visible. Google has dropped its red lines: limits or rules a person or organisation has promised not to cross — 底线 / 红线 around AI weapons. OpenAI has fought a legal battle to shedshed: to get rid of something unwanted; to cause blood, tears, or light to flow or fall — 去除；流（泪）；脱落；散发（光） non-profit status. Anthropic has abandoned its flagshipflagship: The most important one out of a related group. — 旗舰（最重要的一家／一款） policy of never training a model it could not guarantee was safe. Gas-powered datacenterdatacenter: A facility used to house computer systems and associated components, such as telecommunications and storage equipment — 数据中心 campuses are projected to emitemit: to send out or give off — 排放；散发 more greenhouse gas than entire countries, and deepfakesdeepfake: A photo or video created using artificial intelligence, that imposes the face of one person onto the body of another — 深度伪造（影像） have crossed the point of no return: a stage past which it is impossible to undo or reverse a situation — 无法回头的临界点. In almost the same breath as their warnings of apocalypse, the executives promise salvationsalvation: The process of being saved, the state of having been saved (from hell) — 拯救；救赎: a utopia-gifting country of geniuses inside a datacenter, the climate fixed, a colony on another world. Utopia and apocalypse, Vallor argues, are two faces of one mythicmythic: Larger-than-life — 宏大的；传奇般的 coin, pitched on a scale too grand for courts or regulators to feel they have any purchase. "Nothing about them is ungovernableungovernable: Not governable — 无法管理的；不受控制的," she says. "Unless we choose not to govern them."
Questions
Check your understanding
- 01
Which sentence best captures the central argument the writer constructs across the article?
- 02
Why does the writer place the GPT-2 release, the 2023 extinction statement and Musk's launch of xAI in the same paragraph?
- 03
What is the rhetorical effect of Khlaaf's complaint about the missing false positive rate?
- 04
Assess the claim that 'fear-based marketing' is more effective than ordinary product marketing for AI firms. Use at least two examples from the article and one piece of evidence the writer does NOT cite.
Suggested length: ~100 words
- 05
Argue whether the article's closing parallels to the Metaverse, Bitcoin and 'social media saves democracy' strengthen or weaken the writer's case against AI hype.
Suggested length: ~100 words
Questions
Check your understanding
- 01
Which sentence best captures the central argument the writer constructs across the article?
- 02
Why does the writer place the GPT-2 release, the 2023 extinction statement and Musk's launch of xAI in the same paragraph?
- 03
What is the rhetorical effect of Khlaaf's complaint about the missing false positive rate?
- 04
Assess the claim that 'fear-based marketing' is more effective than ordinary product marketing for AI firms. Use at least two examples from the article and one piece of evidence the writer does NOT cite.
Suggested length: ~100 words
- 05
Argue whether the article's closing parallels to the Metaverse, Bitcoin and 'social media saves democracy' strengthen or weaken the writer's case against AI hype.
Suggested length: ~100 words