The article explicitly references AI systems capable of autonomous lethal decision-making in weapons, which are not currently produced by Rheinmetall but could plausibly be developed or used by others. The CEO's call for global regulation acknowledges the risk that such AI weapons could cause harm in the future. Since no actual harm or incident is reported yet, but the potential for harm is credible and significant, this qualifies as an AI Hazard rather than an AI Incident or Complementary Information.[AI generated]
AIM: AI Incidents and Hazards Monitor
Automated monitor of incidents and hazards from public sources (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
Show summary statistics of AI incidents & hazards
Rheinmetall CEO Calls for Ban on Autonomous AI Weapons
Rheinmetall CEO Armin Papperger, at the Eurosatory arms fair near Paris, urged for global regulations to ensure humans, not AI, make lethal decisions in warfare. He warned of rapid advances in military AI and advocated for international rules, similar to those for nuclear weapons, to prevent autonomous AI weapon use.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Autonomy level:
AI system task:
Why's our monitor labelling this an incident or hazard?
AI-Generated Fake Images Target Iranian Footballer, Prompting Legal Action
AI tools were used to create and spread fake images falsely attributed to Iranian footballer Mohammad Mohebi on social media, causing reputational harm. The Iranian Football Federation condemned the images as entirely fabricated, warned against misinformation, and announced potential legal action against those responsible for creating and disseminating the content.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
AI system task:
Why's our monitor labelling this an incident or hazard?
The event involves the use of AI systems to create fake images (deepfakes) that falsely depict a public figure, which constitutes a violation of rights and harm to the individual's reputation (harm to person/community). Since the AI-generated content has already been disseminated and caused harm, this qualifies as an AI Incident. The official denial and warning about the use of AI for misinformation further confirm the AI system's role in causing harm.[AI generated]
Europe Warned of Existential AI Dependency on US Firms
Clemens Fuest, president of Germany's Ifo Institute, warns that Europe's heavy reliance on US AI companies, highlighted by US government restrictions on Anthropic's advanced AI models for foreign nationals, exposes critical sectors like research, industry, and defense to significant risks due to limited European AI infrastructure.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Autonomy level:
AI system task:
Why's our monitor labelling this an incident or hazard?
The article focuses on a strategic and geopolitical concern regarding AI dependency and access restrictions, which could plausibly lead to future harms in critical sectors if dependency issues are not addressed. However, no direct or indirect harm has occurred yet, and no specific AI system malfunction or misuse is described. Therefore, this qualifies as an AI Hazard, reflecting a credible risk of future harm due to AI system access limitations and dependency.[AI generated]
Concerns Over Chinese Military-Linked Researchers at German AI Security Institute
German media revealed that the government-funded CISPA cybersecurity center employs numerous researchers with ties to China's military-industrial complex. Their AI and cybersecurity research, including dual-use technologies, has raised national security concerns about potential technology transfer to China and future misuse, though no actual harm has yet occurred.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Business function:
Why's our monitor labelling this an incident or hazard?
The article explicitly involves AI systems and research related to AI model attacks, misinformation, and critical infrastructure vulnerabilities, which are dual-use technologies with military and civilian applications. The involvement of researchers linked to the Chinese military and intelligence raises credible concerns about the potential misuse or transfer of sensitive AI technologies that could lead to harm. However, the article does not report any actual harm or incident caused by AI systems but warns of plausible future risks. Hence, it fits the definition of an AI Hazard, where the development and use of AI systems could plausibly lead to an AI Incident in the future.[AI generated]
AI Music Generators Face Lawsuits Over Use of Copyrighted Songs
Major AI companies trained music generation models on millions of copyrighted songs without permission, leading to lawsuits from major record labels like Warner Music, Sony Music, and Universal Music Group. Artists and industry leaders allege intellectual property theft and seek compensation for unauthorized use of their works.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Business function:
AI system task:
Why's our monitor labelling this an incident or hazard?
The article explicitly mentions AI systems trained on millions of copyrighted songs without authorization, leading to legal claims of intellectual property rights violations by artists and record companies. The AI systems' development and use have directly led to alleged harm (copyright infringement), triggering lawsuits. This fits the definition of an AI Incident because the AI system's use has directly caused a breach of legal obligations protecting intellectual property rights. The presence of ongoing legal actions further confirms the materialization of harm rather than a mere potential risk.[AI generated]
France Travail's AI Deployment Raises Union Concerns Over Job Security
France Travail is rapidly deploying AI tools like ChatFT, CoachFT, and MatchFT to automate administrative tasks and support employment services across France. While management highlights efficiency gains, unions express concerns about potential job losses and work intensification, demanding safeguards against possible negative impacts of AI integration.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Business function:
AI system task:
Why's our monitor labelling this an incident or hazard?
The event involves AI systems explicitly (ChatFT, ChatFT Écoute, MatchFT) used in employment services. The concerns expressed by unions about job losses and work intensification represent plausible future harms related to AI use. However, no actual harm or incident has occurred yet, only fears and planned negotiations. Therefore, this situation fits the definition of an AI Hazard, as the AI systems' use could plausibly lead to harms such as labor rights violations or work disruption, but these harms are not yet realized.[AI generated]
China Tightens Indium Export Controls, Threatening AI Data Center Supply Chains
China, which produces 70% of global indium, has intensified export scrutiny and imposed controls on indium phosphide, a key material for high-speed optical chips in AI data centers. These measures raise credible risks of future supply chain disruptions for AI infrastructure, though no direct harm has yet occurred.[AI generated]
Industries:
Severity:
Why's our monitor labelling this an incident or hazard?
The event involves AI-related materials (indium for AI data centers), but the article centers on export control policies and trade scrutiny rather than any AI system malfunction, misuse, or harm. There is no direct or indirect harm caused by AI systems reported, only potential future risks related to supply chain constraints. Therefore, this is best classified as an AI Hazard because the increased export controls could plausibly lead to future AI-related harms such as disruption in AI infrastructure development, but no incident has occurred yet.[AI generated]
Meta's AI-Driven Platforms Linked to Scams and Child Harm, Faces Legal Scrutiny
US advocacy groups accuse Meta of allowing AI-driven ad systems to spread scams targeting seniors, causing financial harm. Simultaneously, Meta faces lawsuits over AI-powered content curation linked to child harm on its platforms and is lobbying for immunity from related lawsuits. The incidents highlight AI's role in enabling harm to vulnerable users.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Business function:
Autonomy level:
AI system task:
Why's our monitor labelling this an incident or hazard?
Meta's platforms use AI systems for ad targeting and content recommendation. The fraudulent ads targeting seniors are spread through these AI-driven systems, which have failed to prevent the scams, leading to financial harm to seniors. The event involves the use of AI systems and realized harm (financial scams) to a vulnerable group, meeting the criteria for an AI Incident. The article focuses on the harm caused by the AI system's use rather than just general criticism or policy discussion, so it is not Complementary Information.[AI generated]
AI Companies Accused of Stealing Australian and New Zealand Music for Training Datasets
Multiple investigations reveal AI companies have used millions of copyrighted songs by Australian and New Zealand artists, including Kylie Minogue and Lorde, to train generative AI models without permission or payment. This unauthorized use has led to significant financial and cultural harm to creators, sparking industry backlash.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Business function:
Autonomy level:
AI system task:
Why's our monitor labelling this an incident or hazard?
The event explicitly involves AI systems being developed using large datasets containing copyrighted music without authorization, leading to a violation of intellectual property rights and financial harm to artists. This fits the definition of an AI Incident because the AI system's development and use have directly led to a breach of intellectual property rights and economic harm to individuals (artists). The article details the harm occurring and the lack of licensing or payment, confirming realized harm rather than potential harm.[AI generated]
Western Australia Police Deploy Real-Time AI Facial Recognition Cameras
Western Australia Police have begun deploying real-time AI-powered facial recognition cameras in public spaces to identify suspects and offenders. The technology, mounted on police vehicles, cross-references faces with a database. Critics raise concerns about privacy, potential misidentification, and lack of public consultation, highlighting risks of rights violations.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Business function:
Autonomy level:
AI system task:
Why's our monitor labelling this an incident or hazard?
The article explicitly involves an AI system (real-time facial recognition) used by police. The system is currently being trialed and operational, but no direct or indirect harm has been reported. The article discusses potential risks and safeguards, indicating plausible future harm such as privacy violations or wrongful arrests. Since no harm has occurred yet, but the AI system's use could plausibly lead to an AI Incident, the event is best classified as an AI Hazard rather than an AI Incident or Complementary Information. It is not unrelated because the AI system is central to the event.[AI generated]
AI-Generated Medical Leaflet Contains Offensive Language, Causing Patient Harm in South Korea
A South Korean doctor used an AI system to generate a patient guidance document, but the AI prompt—containing demeaning language about the patient's intelligence and occupation—was printed and handed directly to the patient. The incident sparked public outrage over the violation of patient dignity and the misuse of AI in healthcare.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Business function:
Autonomy level:
AI system task:
Why's our monitor labelling this an incident or hazard?
The event clearly involves an AI system since the doctor used an AI prompt to generate a patient guidance document. The harm is realized as the patient was directly subjected to dehumanizing and insulting language, which is a violation of human rights and dignity. The AI system's role is pivotal because the harmful prompt was part of the AI input and was mistakenly included in the output given to the patient. This is a direct harm caused by the AI system's use and the human error in handling the AI output, fitting the definition of an AI Incident.[AI generated]
AI-Generated Deepfake Video Sparks Legal Action in Brazilian Election
The Federation Brasil da Esperança (PT, PCdoB, PV) filed a complaint with Brazil's electoral court against Senator Flávio Bolsonaro for distributing an AI-generated deepfake video. The video parodied Top Gun, associating the PT with organized crime, raising concerns about electoral misinformation and the integrity of the democratic process.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Autonomy level:
AI system task:
Why's our monitor labelling this an incident or hazard?
The event describes the use of AI-generated deepfake technology to create misleading political content that harms the integrity of the electoral process and political rights. This constitutes a violation of legal protections and human rights related to fair political participation. Since the AI system's use has directly led to this harm and legal action, it qualifies as an AI Incident under the framework.[AI generated]
Germany and Ukraine Launch Large-Scale Production of AI-Enabled Combat Robots for Ukraine
Quantum Tencore Industries, a German-Ukrainian joint venture, will produce 2,000 AI-enabled Termit unmanned ground vehicles for the Ukrainian Armed Forces. Funded by Germany's Ministry of Defense, these battle-tested robots, used in logistics and combat, mark Europe's largest UGV procurement and are actively deployed in Ukraine's conflict zone.[AI generated]
AI principles:
Industries:
Severity:
Business function:
AI system task:
Why's our monitor labelling this an incident or hazard?
The TerMIT unmanned ground vehicles are AI systems as they perform complex autonomous or semi-autonomous tasks in combat and support roles, including logistics and casualty evacuation, which influence physical environments and human safety. Their deployment in active combat zones has already led to battlefield impacts, implying direct or indirect harm to persons and communities. The article explicitly states these robots have been battle-tested and are actively used in missions that affect human lives, fulfilling the criteria for an AI Incident. Although the article focuses on production and scaling, the operational use and battlefield effects are central, indicating realized harm rather than just potential risk.[AI generated]
Argentina Prepares First Autonomous Vehicle Corridor Amid Regulatory Changes
Argentina's government, led by Minister Federico Sturzenegger, has modified regulations to allow autonomous vehicles on National Route 14, aiming to reduce road accidents and attract companies like Tesla. The initiative marks the country's first corridor for AI-driven vehicles, raising future safety and regulatory considerations.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Autonomy level:
AI system task:
Why's our monitor labelling this an incident or hazard?
The article explicitly mentions AI systems in the form of autonomous driving technology and government regulatory changes to permit their use. While the article highlights the potential safety benefits of autonomous vehicles, it does not describe any actual harm, accident, or malfunction caused by AI. The focus is on the government's preparations and optimism about AI applications, including autonomous driving, which could plausibly lead to incidents if issues arise during deployment. Since no harm has yet occurred, and the event concerns future deployment and regulatory framework, it fits the definition of an AI Hazard. It is not Complementary Information because it is not an update or response to a past incident, nor is it unrelated as it clearly involves AI systems and their societal implications.[AI generated]
Spanish Companies Unprepared for AI-Driven Cyberattacks
A recent study reveals that 65% of Spanish organizations lack formal plans to assess or respond to AI-powered cyberattacks, such as advanced phishing schemes. This lack of preparedness increases the risk of harm from evolving AI-enhanced threats, highlighting a significant cybersecurity vulnerability in Spain.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
AI system task:
Why's our monitor labelling this an incident or hazard?
The article centers on the potential for AI-enhanced cyberattacks (like phishing 2.0) to cause harm in the future due to insufficient preparedness by Spanish companies. No actual AI incident causing harm is reported; rather, the article warns about plausible future harms from AI-powered cyber threats. Therefore, this qualifies as an AI Hazard because it describes credible risks that could plausibly lead to AI incidents if not addressed. It is not Complementary Information since it is not updating or responding to a past incident, nor is it unrelated as it clearly involves AI and cybersecurity risks.[AI generated]
AI-Generated Deepfake Portrays German Football Fan as Hitler, Causing Harm
An AI-generated deepfake image falsely depicting German football fan Jan Weitzel as Adolf Hitler circulated widely on social media after a World Cup match in Houston. The manipulated image caused significant reputational and emotional harm to Weitzel and his family, prompting legal action and raising concerns about AI misuse for defamation.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
AI system task:
Why's our monitor labelling this an incident or hazard?
The event involves the use of AI systems (OpenAI tools) to generate manipulated visual content that falsely portrays a person in a harmful and defamatory manner. The harm is realized as the individual suffers reputational damage and distress, prompting legal action. The AI system's role is pivotal in creating and spreading the manipulated content. This meets the criteria for an AI Incident due to direct harm to a person's rights and reputation caused by AI-generated misinformation.[AI generated]
AI Facial Recognition at ATMs Reduces Fraud in Taiwan
Taiwanese police and financial institutions deployed AI-powered facial recognition and risk detection systems at ATMs to identify users with obscured faces, a common tactic by fraudsters. This initiative led to a significant drop in fraud cases and financial losses, effectively disrupting criminal cash withdrawals and reducing harm to property and communities.[AI generated]
Industries:
Severity:
Business function:
Autonomy level:
AI system task:
Why's our monitor labelling this an incident or hazard?
The event involves the use of an AI system (facial recognition with AI computer vision) in ATM fraud prevention. The AI system's use has directly led to a significant reduction in financial losses from fraud, which constitutes harm to property and communities. Since the AI system's deployment has already caused a measurable decrease in harm, this qualifies as an AI Incident rather than a hazard or complementary information. The article focuses on the AI system's role in preventing harm, not just potential or future risks, nor is it merely an update or general news.[AI generated]
South Korean Actress Lee Ji-hye Targeted by Deepfake AI Scam Ads
South Korean actress Lee Ji-hye was targeted by deepfake AI technology, with her likeness and voice used in fake online ads promoting products like lingerie and sweet potatoes. The ads, created by Chinese entities, misled consumers and violated her rights, prompting her to publicly warn fans against the scam.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Business function:
AI system task:
Why's our monitor labelling this an incident or hazard?
The event explicitly involves AI-generated synthetic media (deepfake videos) that impersonate a real person to promote products fraudulently. This misuse of AI has directly led to harm in the form of reputational damage and consumer deception, which can be considered harm to communities and a violation of personal rights. Therefore, this qualifies as an AI Incident under the framework.[AI generated]
Morocco Partners with Harmattan AI to Develop Autonomous Defense Systems
The Moroccan Armed Forces have entered a strategic partnership with Harmattan AI to develop, produce, and deploy AI-powered autonomous defense systems, including anti-drone technologies. The agreement includes establishing local manufacturing, research centers, and academic collaborations, raising credible future risks associated with large-scale deployment of autonomous military AI systems.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Autonomy level:
AI system task:
Why's our monitor labelling this an incident or hazard?
The article explicitly mentions the development and production of autonomous defense systems, which are AI systems by definition. Although no incident or harm has been reported, the creation and proliferation of autonomous weapons systems are widely recognized as potential sources of serious harm, including injury, human rights violations, and destabilization. The event focuses on the establishment of capabilities and partnerships for such AI-enabled defense technologies, indicating a credible risk of future harm. Hence, it fits the definition of an AI Hazard rather than an Incident or Complementary Information.[AI generated]
Graham Norton Wins US Court Order Against Meta Over Harmful AI Deepfakes
Irish broadcaster Graham Norton won a US court order compelling Meta to reveal the operators behind a Facebook page spreading AI-generated deepfake images and false claims about him and his family. The posts caused significant distress and reputational harm, prompting legal action to identify and pursue those responsible.[AI generated]
AI principles:
Industries:
Affected stakeholders:
Harm types:
Severity:
Autonomy level:
AI system task:
Why's our monitor labelling this an incident or hazard?
The event explicitly mentions AI-generated images and fabricated stories that have caused significant alarm, distress, and anxiety to Graham Norton, which qualifies as harm to a person. The AI system's outputs (deepfakes and false posts) directly led to this harm. The involvement of Meta (Facebook) as the platform where these AI-generated harmful contents were posted further confirms the AI system's role. Hence, this is an AI Incident as per the definitions provided.[AI generated]