教育与科研
中小学、大学和研究机构：AI助长的作弊造假、针对学生的深度伪造侵害以及研究数据泄露。
最近 12 个月 · 自 2025年11月4日 起
官方警告与应对准备
政府首脑、部长、监管机构、央行、政府间组织和国家网络安全机构发布的公开警告、演练、指引和新规。每条均附来源链接。
- 指引CISA, NSA and FBI · 安全机构美国
US agencies warn of industrial-scale knowledge distillation campaigns against US AI models
The agencies said China-based AI companies have extracted billions of tokens from US frontier models since at least late 2024, and urged AI developers to detect suspicious activity, alter responses to suspected distillation and share intelligence.
来源: CISA
- 指引eSafety Commissioner (Australia) · 监管机构澳大利亚
Australia's eSafety Commissioner issues advisory on AI misuse of school photos
eSafety urged schools to review how they share images online after receiving over 100 reports in January to March 2026 of accounts misusing school photos, including AI-generated sexualised deepfakes and face swaps of students and staff.
来源: Human Rights Watch
- 规则UK Department for Education · 部长英国
Keeping Children Safe in Education 2026 covers AI-generated nude images
The 2026 statutory safeguarding guidance, in force from 1 September 2026, defines sharing nudes and semi-nudes to include images digitally altered or wholly generated using AI, and links schools to generative AI safety expectations.
来源: EdTech Innovation Hub
- 规则Minnesota legislature · 立法机构美国
Minnesota passes first US ban on 'nudification' apps
According to 19th News, Minnesota has passed what it describes as the nation's first ban on 'nudification' apps, which use AI to create fake nude images of real people.
来源: 19thnews.org via Hacker News
- 警告UK National Crime Agency and Early Warning Working Group · 安全机构英国
UK schools advised to remove identifiable pupil photos as AI blackmail threat grows
The National Crime Agency warned that criminals use AI to alter children's photos taken from school sites and demand payment, and schools were advised to remove or limit identifiable pupil images online.
来源: Internet Watch Foundation (republishing The Guardian)
- 警告White House Office of Science and Technology Policy (Director Michael Kratsios) · 政府首脑美国
White House memo NSTM-4 addresses adversarial distillation of US AI models
The memo said foreign entities, principally in China, run industrial-scale campaigns to distil US frontier AI models, and committed agencies to share information with AI developers and explore accountability measures.
来源: The White House
- 指引Massachusetts Department of Elementary and Secondary Education (Healey-Driscoll administration) · 监管机构美国
Massachusetts tells school districts to promptly investigate AI deepfake images of students
A letter to districts statewide said creating a sexual deepfake of a minor is a crime and directed school leaders to investigate complaints about AI-generated explicit images of students promptly.
来源: The Boston Globe
- 指引UK Department for Education · 部长英国
UK Department for Education updates generative AI product safety standards for schools
The update added standards on cognitive, emotional and social development, mental health and protection against manipulation to the capabilities generative AI products used in schools and colleges should meet.
来源: GOV.UK
警告以英文摘要呈现。
已核实事件
+13
超过 12 个月（存档）: 41
该行业如何遭受攻击
过去12个月该行业案例档案中最常见的攻击类型，以及其中提及的AI产品。
提及的AI产品
快讯最新报道
- La Voz reports AI was used in a mass cyberattack on a UK school network, matching the 3AM ransomware case already tracked. La Voz del Interior
- Bitdefender describes human-AI hybrid tactics in a September 3AM ransomware attack on a UK school estate that exfiltrated financial data. Infosecurity Magazine
- Bitdefender says a 3AM ransomware affiliate breached a UK school network in a suspected AI-assisted hybrid attack. GBHackers
- arXiv paper presents an agentic red-teaming method for resource-exhaustion attacks on FEM solvers. arXiv cs.CR
- Hackers reportedly used AI agents in 3AM ransomware attacks on UK school networks. CyberSecurityNews
- 3AM ransomware reportedly hit 37 UK school systems in a suspected AI-assisted attack. Hackread
- A 3AM ransomware affiliate encrypted about 37 of nearly 50 machines at a UK school estate in a suspected AI-assisted attack. BleepingComputer
- Researchers say a model may have generated and revised commands in a 3AM ransomware attack on a European school, without claiming full automation. Dark Reading