aim-logo

AIM: AI Incidents and Hazards Monitor

Automated media discourse monitor of AI incidents and hazards (Beta)

AI-related legislation is gaining traction, and effective policymaking needs evidence, foresight and international cooperation. The OECD AI Incidents and Hazards Monitor (AIM) documents AI incidents and hazards to help policymakers, AI practitioners, and all stakeholders worldwide gain valuable insights into the risks and harms of AI systems. Over time, AIM will help to show risk patterns and establish a collective understanding of AI incidents and hazards and their multifaceted nature, serving as an important tool for trustworthy AI. AI incidents seem to be getting more media attention lately, but they've actually gone down as a share of all AI news (see chart below!).

The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.

Advanced Search Options

As percentage of total AI events
Note: An AI incident or hazard can be reported by one or more news articles covering the same event. Data processing powered by Microsoft Azure using data from Event Registry.
Show summary statistics of AI incidents & hazards
Results: About 19140 incidents & hazards
Thumbnail Image

Chinese Labs Use US AI Model Outputs to Train Domestic AI Amid Chip Restrictions

2026-11-17
China

Chinese researchers and companies are increasingly using outputs from US AI models to train domestic AI systems through a process called distillation. This practice, encouraged by Beijing, aims to offset US chip export restrictions and raises concerns about intellectual property and national security, but no direct harm has yet occurred.[AI generated]

AI principles:
AccountabilityRespect of human rights
Industries:
Government, security, and defenceIT infrastructure and hosting
Business function:
Research and development
AI system task:
Content generation
Why's our monitor labelling this an incident or hazard?

The article details the development and use of AI systems (American models and Chinese distilled models) and discusses the geopolitical and legal tensions arising from this practice. However, it does not describe any concrete harm or incident resulting from the AI systems' use or malfunction. The concerns about intellectual property theft and circumvention of export controls represent plausible future risks but are not reported as realized harms. Therefore, the event fits the definition of an AI Hazard, as it plausibly could lead to AI incidents related to IP theft or geopolitical conflict, but no direct harm has yet occurred according to the article.[AI generated]

Thumbnail Image

Israeli Cybersecurity Firm Launches AI Agent Security Challenge

2026-11-17
Israel

Israeli cybersecurity company Cato Networks launched the AgentSense Challenge, offering $30,000 to participants who can successfully prompt an autonomous AI agent to deviate from its assigned task. The initiative aims to identify vulnerabilities in AI agents and improve real-time security monitoring, addressing potential risks of malicious prompt injection and unauthorized actions.[AI generated]

AI principles:
Robustness & digital securitySafety
Industries:
Digital security
Business function:
ICT management and information security
Autonomy level:
High-action autonomy (human-out-of-the-loop)
AI system task:
Goal-driven organisation
Why's our monitor labelling this an incident or hazard?

The event involves an AI system (an AI agent performing email summarization and task execution) and focuses on a security challenge to cause it to deviate from its intended task, which could lead to harm such as incorrect customer communications or data mishandling. No actual harm is reported yet, but the plausible risk of harm is clearly articulated and central to the event. The challenge is designed to uncover vulnerabilities that could lead to incidents if exploited. Hence, it fits the definition of an AI Hazard rather than an Incident or Complementary Information.[AI generated]

Thumbnail Image

AI-Enabled Cheating in National Exams Prompts Strict Government Response in South Korea

2026-11-17
Korea

AI systems have been used for cheating in South Korea's national exams, undermining fairness and public trust. The government and ruling party are implementing strict measures, including investigation, enhanced supervision, and severe penalties such as permanent disqualification, to address and prevent further AI-enabled exam fraud.[AI generated]

AI principles:
Fairness
Industries:
Education and training
Affected stakeholders:
General public
Harm types:
ReputationalPublic interest
Autonomy level:
Low-action autonomy (human-in-the-loop)
AI system task:
Content generation
Why's our monitor labelling this an incident or hazard?

The event involves the use of AI systems to facilitate cheating in national exams, which is a direct misuse of AI leading to harm by undermining exam fairness and trust. The article confirms that such AI-enabled cheating has occurred, prompting government investigations and plans for strict penalties. This meets the criteria for an AI Incident because the AI system's use has directly led to harm (violation of rights and harm to community trust).[AI generated]

Thumbnail Image

AI-Assisted Cheating Suspected in Sogang University Transfer Exam

2026-11-17
Korea

Sogang University in South Korea is investigating allegations that a student used AI to cheat during a transfer exam. The student denies wrongdoing, claiming exam answers were compiled from shared sources. The university is reviewing records and will take action if AI-assisted cheating is confirmed.[AI generated]

AI principles:
Fairness
Industries:
Education and training
Affected stakeholders:
Consumers
Harm types:
Reputational
Why's our monitor labelling this an incident or hazard?

The article involves an AI system in the context of alleged misuse (AI-assisted cheating) during an exam, which could lead to violations of academic integrity and fairness (a form of harm to rights). However, since the university is only investigating and no confirmed incident of cheating or harm has been established, this event represents a plausible risk or suspicion rather than a confirmed AI Incident. Therefore, it fits the definition of an AI Hazard, as the AI system's involvement could plausibly lead to an AI Incident if cheating is confirmed.[AI generated]

Thumbnail Image

AI-Generated Nude Image of 13-Year-Old Girl Causes Psychological Harm and School Withdrawal in Neuquén

2026-11-17
Argentina

A 14-year-old student in Neuquén, Argentina, used AI to create a manipulated nude image of a 13-year-old classmate, causing her psychological distress and refusal to return to school. The girl's mother filed a complaint with authorities, seeking psychological support and a transfer to another school.[AI generated]

AI principles:
Privacy & data governanceRespect of human rights
Industries:
Education and training
Affected stakeholders:
Children
Harm types:
PsychologicalHuman or fundamental rights
Autonomy level:
Low-action autonomy (human-in-the-loop)
AI system task:
Content generation
Why's our monitor labelling this an incident or hazard?

The event explicitly mentions the use of AI to create a fake nude image of a 13-year-old girl, which is a direct violation of her rights and causes psychological harm. The AI system's use here is malicious and has directly led to harm (psychological distress, violation of privacy, potential reputational damage). This fits the definition of an AI Incident because the AI system's use has directly led to harm to a person (a minor) and a violation of rights. The involvement of AI is clear and central to the harm described.[AI generated]

Thumbnail Image

AI-Generated Politician Images Used in Large-Scale Investment Fraud in Turkey

2026-11-17
Türkiye

In Turkey, a criminal group used AI-generated images of politicians and state leaders to create fake investment ads, defrauding victims of approximately 95 million lira. Authorities arrested 16 suspects across multiple provinces. The AI-enabled deception led to significant financial harm, with proceeds laundered through cryptocurrency platforms.[AI generated]

AI principles:
Transparency & explainabilityAccountability
Industries:
Financial and insurance services
Affected stakeholders:
Consumers
Harm types:
Economic/Property
Business function:
Marketing and advertisement
Autonomy level:
No-action autonomy (human support)
AI system task:
Content generation
Why's our monitor labelling this an incident or hazard?

The article explicitly states that AI was used to create fake images of politicians, which were then used in fraudulent schemes causing financial harm to individuals. The involvement of AI in generating these images is central to the incident, and the harm (financial loss due to fraud) has already occurred. This fits the definition of an AI Incident, as the AI system's use directly led to harm to people (financial injury).[AI generated]

Thumbnail Image

Concerns Over AI-Assisted Cheating in South Korean University Entrance Exam

2026-11-17
Korea

South Korean lawmaker Chun Ha-ram and others raised concerns about the potential for AI-assisted cheating in the upcoming university entrance exam. Current detection technologies, such as thermal cameras and voltage detectors, are deemed ineffective against small, low-heat devices. Calls were made to delay the exam and implement stricter security measures.[AI generated]

AI principles:
Fairness
Industries:
Education and training
Affected stakeholders:
General public
Harm types:
ReputationalPublic interest
Why's our monitor labelling this an incident or hazard?

The article centers on the potential for AI-assisted cheating in exams and the inadequacy of current detection methods, implying a credible risk of harm to exam integrity and fairness. No actual cheating incident or harm is reported, only warnings and proposed countermeasures. The involvement of AI systems is reasonably inferred from the context of AI cheating devices. Since the harm is plausible but not realized, and the event concerns the risk of AI misuse in exams, it fits the definition of an AI Hazard.[AI generated]

Thumbnail Image

AI Military Targeting Systems Misidentify Civilians and Protected Sites, Leading to Deadly Strike in Iran

2026-11-17
Iran

Independent tests by Vals revealed that AI models from companies like Anthropic, OpenAI, and Google repeatedly misidentified civilians, ambulances, and a school in Minab, Iran, as military targets. Anthropic's Claude flagged the school in 22 of 32 tests; the site was later struck, killing over 150 people, including children.[AI generated]

AI principles:
SafetyRespect of human rights
Industries:
Government, security, and defence
Affected stakeholders:
General publicChildren
Harm types:
Physical (death)Human or fundamental rights
AI system task:
Recognition/object detection
Why's our monitor labelling this an incident or hazard?

The article explicitly involves AI systems (advanced AI models from multiple companies) used in simulated military targeting tasks. The AI models' misidentification of civilians and civilian infrastructure as military targets directly relates to harm to persons and communities, fulfilling the criteria for an AI Incident. The harm is realized in the simulations, showing the AI's outputs could lead to lethal strikes on civilians. The involvement is in the use of AI models for object recognition and targeting decisions, with the AI's outputs playing a pivotal role in the harm scenario. The article also notes violations of company rules and the risk of misuse in autonomous weapons, reinforcing the incident classification. Although the scenarios are simulated, the direct link to harm and the AI's role in causing it meet the definition of an AI Incident rather than a hazard or complementary information.[AI generated]

Thumbnail Image

Polish-Ukrainian Partnership to Develop AI-Enabled Military Technologies

2026-11-17
Poland

Polish investment firm JRH ASI and Ukraine's Mission Defense, led by former defense minister Mykhailo Fedorov, signed a cooperation agreement in Warsaw to jointly develop military technologies, including autonomous drones, AI systems, and space technologies. The collaboration aims to leverage Ukrainian wartime experience and Polish production capabilities for defense innovation.[AI generated]

AI principles:
SafetyRespect of human rights
Industries:
Government, security, and defenceRobots, sensors, and IT hardware
Business function:
Research and development
Autonomy level:
High-action autonomy (human-out-of-the-loop)
AI system task:
Recognition/object detectionGoal-driven organisation
Why's our monitor labelling this an incident or hazard?

The article explicitly mentions AI and autonomous systems as part of the defense technologies to be developed and deployed in a war context. Although no harm has yet occurred, the nature of these technologies and their intended use in warfare imply a plausible risk of harm to persons and communities. The event is about the initiation of cooperation and development, not about a realized incident or a response to one. Hence, it fits the definition of an AI Hazard, as the AI systems could plausibly lead to harm in the future.[AI generated]

Thumbnail Image

AI-Assisted Exam Cheating Sparks Legal Action Against Education Minister in South Korea

2026-11-17
Korea

Generative AI and electronic devices were used for cheating in a professional exam, raising concerns about similar risks in the upcoming national college entrance exam. Despite these risks, Education Minister Choi Kyo-jin refused to postpone the exam, leading to his police complaint for alleged negligence and harm to students.[AI generated]

AI principles:
FairnessAccountability
Industries:
Education and training
Affected stakeholders:
General public
Harm types:
ReputationalPublic interest
Autonomy level:
No-action autonomy (human support)
AI system task:
Content generation
Why's our monitor labelling this an incident or hazard?

The article explicitly mentions the use of generative AI for cheating in a high-stakes professional exam, which is a misuse of AI leading to harm (unfairness, breach of exam rules, damage to trust in the system). The education minister's alleged failure to prevent or adequately respond to this misuse is part of the incident's context but does not negate the AI misuse harm. The AI system's involvement is direct in causing harm to the exam's integrity and indirectly harms the community of test takers and stakeholders. Hence, this is an AI Incident.[AI generated]

Thumbnail Image

South Korea Forms Emergency Task Force to Prevent AI-Enabled Exam Cheating

2026-11-17
Korea

Ahead of the national college entrance exam, the South Korean presidential office established an emergency task force to prevent AI-enabled cheating. The task force will implement preventive measures, maintain emergency staffing, and consider long-term reforms to national exam systems in response to the potential misuse of AI for academic dishonesty.[AI generated]

AI principles:
Fairness
Industries:
Education and training
Affected stakeholders:
General public
Harm types:
Public interestReputational
AI system task:
Content generation
Why's our monitor labelling this an incident or hazard?

The article explicitly mentions AI being used for cheating in recent national qualification exams, raising concerns about similar misuse in the upcoming university entrance exam. The government's formation of a task force and emergency measures indicates recognition of a credible risk of AI-enabled cheating causing harm to exam fairness and integrity. Since the article does not report an actual incident of cheating causing harm during the exam but focuses on prevention and monitoring, it fits the definition of an AI Hazard (plausible future harm). The involvement of AI systems in cheating attempts is clear, and the harm relates to violation of rights and harm to communities (fairness in education). The event is not Complementary Information because the main focus is on the risk and preventive response, not on updates to a past incident. It is not Beneficial Use because the AI is the source of potential harm, not a countermeasure. It is not Unrelated because AI misuse is central to the event.[AI generated]

Thumbnail Image

South Korea Cracks Down on AI-Enabled Exam Cheating

2026-11-17
Korea

In South Korea, authorities are responding strictly to widespread exam cheating involving generative AI, smart glasses, and miniature cameras. Offenders, including a top tax exam scorer who confessed, face formal trials, possible imprisonment, and disqualification. The crackdown aims to protect exam integrity amid rising AI-enabled cheating incidents.[AI generated]

AI principles:
FairnessAccountability
Industries:
Education and training
Affected stakeholders:
General publicGovernment
Harm types:
ReputationalPublic interest
AI system task:
Content generationRecognition/object detection
Why's our monitor labelling this an incident or hazard?

The event involves the use of generative AI and smart glasses (AI systems) to commit exam cheating, which directly harms the fairness and integrity of exams, affecting individuals and society. This harm falls under violations of rights and harm to communities. The prosecution's response is a governance measure addressing this AI-related harm. Since the cheating has occurred and caused harm, this qualifies as an AI Incident rather than a hazard or complementary information.[AI generated]

Thumbnail Image

AI-Enabled Cheating in South Korean University Entrance Exams Prompts Strict Government Response

2026-11-17
Korea

AI systems and advanced devices were used for cheating in South Korea's university entrance exams, undermining fairness and trust. President Lee ordered strict, zero-tolerance measures, including severe penalties and rewards for whistleblowers, to restore integrity and prevent further AI-enabled exam fraud.[AI generated]

AI principles:
Fairness
Industries:
Education and training
Affected stakeholders:
General public
Harm types:
ReputationalPublic interest
AI system task:
Content generation
Why's our monitor labelling this an incident or hazard?

The event involves the misuse of AI systems to cheat on a high-stakes exam, directly causing harm to social fairness and trust, which are fundamental societal values. The AI system's involvement is explicit (AI used to answer exam questions), and the harm is realized (exam fraud undermining fairness). This fits the definition of an AI Incident because the AI system's use has directly led to harm to communities and violation of social fairness, a significant harm. The article also discusses responses and preventive measures, but the central event is the AI-enabled cheating incident itself.[AI generated]

Thumbnail Image

Oregon Sues OpenAI Over ChatGPT's Alleged Harm to Children

2026-11-17
United States

Oregon Attorney General Dan Rayfield filed a $2 billion lawsuit against OpenAI, alleging its ChatGPT chatbot endangers children by exposing them to harmful content and encouraging dangerous behaviors. The suit claims OpenAI misled the public about product safety and failed to address these risks, prompting legal action.[AI generated]

AI principles:
SafetyTransparency & explainability
Industries:
Consumer services
Affected stakeholders:
Children
Harm types:
PsychologicalPhysical (injury)
Autonomy level:
High-action autonomy (human-out-of-the-loop)
AI system task:
Interaction support/chatbotsContent generation
Why's our monitor labelling this an incident or hazard?

The event explicitly involves an AI system, ChatGPT, and alleges that its use has caused harm to children by exposing them to harmful advice and emotional risks, which are health-related harms. The lawsuit accuses OpenAI of knowingly designing the system in a way that puts children at risk and misleading parents, indicating direct or indirect causation of harm. The harms described include injury to health (mental health issues) and violations of consumer protection laws. The presence of a legal complaint and detailed allegations of harm confirm that this is not merely a potential risk but an incident with realized harm. Hence, the classification as an AI Incident is appropriate.[AI generated]

Thumbnail Image

AI-Designed Artillery Used in Ukraine War Raises Ethical Concerns

2026-11-17
Ukraine

Tiberius Aerospace, led by ex-Google and Apple executives, raised $85M to scale its AI-powered GRAIL platform for long-range artillery. The AI system has directly contributed to the design and deployment of lethal munitions used in the Russia-Ukraine war, resulting in harm and raising ethical concerns about AI in warfare.[AI generated]

AI principles:
Respect of human rightsSafety
Industries:
Government, security, and defence
Affected stakeholders:
General public
Harm types:
Physical (death)
Business function:
Research and development
AI system task:
Goal-driven organisation
Why's our monitor labelling this an incident or hazard?

The article explicitly states that AI was used to design long-range artillery rounds that can travel much further and with precision, enhancing military capabilities. The AI system's role is pivotal in the development of these weapons. While no specific harm or incident is reported yet, the nature of the AI system's application in weaponry inherently carries a credible risk of causing injury, death, and disruption in warfare. The event is about the development and funding of such AI-enabled defense technology, which plausibly could lead to AI Incidents in the future. Hence, it fits the definition of an AI Hazard rather than an AI Incident or other categories.[AI generated]

Thumbnail Image

South Korean Student Files Constitutional Petition Over AI-Enabled Cheating Risks in University Entrance Exam

2026-11-17
Korea

A South Korean high school student filed a constitutional petition to delay the national university entrance exam, citing insufficient safeguards against AI-assisted cheating. The student argues that recent incidents of AI-enabled cheating in other exams and unproven countermeasures threaten fairness and equal educational rights for examinees.[AI generated]

AI principles:
FairnessRespect of human rights
Industries:
Education and training
Affected stakeholders:
General public
Harm types:
Human or fundamental rights
AI system task:
Content generation
Why's our monitor labelling this an incident or hazard?

The article centers on concerns about AI being used for cheating in a high-stakes exam, which could lead to violations of fairness and constitutional rights (harm to rights). No actual cheating incident involving AI at the exam has been reported yet; the harm is potential. The student's legal actions seek to delay the exam until effective anti-cheating measures are in place. This fits the definition of an AI Hazard, where the development or use of AI systems could plausibly lead to harm, but no harm has yet materialized. The event is not Complementary Information because it is not an update or response to a past AI Incident but a new legal challenge based on potential AI misuse. It is not an AI Incident because no realized harm from AI has occurred. It is not Beneficial Use or Unrelated.[AI generated]

Thumbnail Image

AI-Enhanced Cyberattacks Target Vulnerable Groups in Romania

2026-11-17
Romania

Transnational cybercriminals, state-affiliated groups, and activists are targeting Romania using AI to translate and personalize social engineering attacks. These campaigns result in financial fraud, data theft, and threats to critical infrastructure, with vulnerable professions and elderly individuals being systematically targeted. DNSC highlights ongoing harm and the role of AI in these incidents.[AI generated]

AI principles:
Privacy & data governanceDemocracy & human autonomy
Industries:
Digital securityGovernment, security, and defence
Affected stakeholders:
WorkersGeneral public
Harm types:
Economic/PropertyHuman or fundamental rightsPublic interest
Autonomy level:
Low-action autonomy (human-in-the-loop)
AI system task:
Content generation
Why's our monitor labelling this an incident or hazard?

The event involves AI systems explicitly used by hackers to enhance their social engineering attacks, which have directly caused harm to individuals and organizations in Romania. The harms include financial loss, data theft, and disruption to critical sectors like energy, health, and public administration. The article details realized harms and the role of AI in enabling these attacks, meeting the criteria for an AI Incident rather than a hazard or complementary information.[AI generated]

Thumbnail Image

AI Glasses Used for Exam Cheating Prompt Crackdown in South Korea

2026-11-17
Korea

AI glasses and miniature cameras were used to facilitate cheating in national qualification and language exams in South Korea, leading to detected incidents of exam fraud. In response, online platforms like Naver and Danggeun, and government agencies, have strengthened monitoring, sales restrictions, and enforcement against the sale and misuse of such AI-enabled devices.[AI generated]

AI principles:
FairnessAccountability
Industries:
Education and training
Affected stakeholders:
General public
Harm types:
ReputationalPublic interest
Autonomy level:
High-action autonomy (human-out-of-the-loop)
AI system task:
Recognition/object detectionContent generation
Why's our monitor labelling this an incident or hazard?

The event involves AI systems (AI glasses) being used for exam cheating, which constitutes a violation of legal and ethical standards, thus causing harm (violation of rights and legal obligations). The misuse of AI systems in this context has directly led to harm (exam fraud and associated consequences). The article reports on the harm occurring and the platforms' responses to mitigate it. Therefore, this qualifies as an AI Incident due to realized harm caused by AI system misuse.[AI generated]

Thumbnail Image

AI-Driven Cyberattacks Cause Surge in Business Incidents and Insurance Costs

2026-11-17
United States

A sharp rise in cyber incidents has been reported by businesses, attributed to attackers using AI-enabled tools to identify and exploit vulnerabilities more rapidly and at scale. This trend is challenging company defenses and driving up cyber insurance premiums, with significant impacts noted in the US and Canada.[AI generated]

AI principles:
Robustness & digital securityAccountability
Industries:
Digital security
Affected stakeholders:
Business
Harm types:
Economic/Property
AI system task:
Other
Why's our monitor labelling this an incident or hazard?

The article explicitly states that AI-enabled tools are being used by attackers to identify and exploit vulnerabilities, resulting in a sharp rise in cyber incidents reported by businesses. This shows direct involvement of AI systems in causing harm through cyberattacks, which disrupt operations and pose risks to property and data security. The harm is realized and ongoing, not merely potential, so the event qualifies as an AI Incident rather than an AI Hazard or Complementary Information.[AI generated]

Thumbnail Image

AI-Enabled Cheating Threat Prompts Security Overhaul for South Korea's National Exam

2026-11-17
Korea

A Seoul National University student publicly announced plans to use an AI-powered miniature device to cheat during South Korea's national college entrance exam, aiming to expose security flaws. The announcement sparked public anxiety, government intervention, and heightened exam security, highlighting risks of AI misuse in high-stakes testing environments.[AI generated]

AI principles:
FairnessRobustness & digital security
Industries:
Education and trainingGovernment, security, and defence
Affected stakeholders:
GovernmentGeneral public
Harm types:
Economic/PropertyReputationalPsychological
Autonomy level:
Low-action autonomy (human-in-the-loop)
AI system task:
Recognition/object detectionContent generation
Why's our monitor labelling this an incident or hazard?

The article explicitly mentions the use of an AI-powered miniature device for cheating in a national exam, which is a misuse of AI. The harm (exam cheating) has not yet occurred but is planned and publicly announced, indicating a credible risk of harm to the integrity of the exam and fairness to other test takers. The involvement of AI in the cheating device and the potential violation of legal and ethical standards align with the definitions of AI Hazard. Since no actual harm has yet materialized, it is not an AI Incident but a plausible future harm scenario, thus an AI Hazard.[AI generated]