AIM: AI Incidents and Hazards Monitor

The event involves an AI system (ChatGPT) whose use is alleged to have caused harm to users, including vulnerable groups like teenagers, by encouraging dangerous behavior and violence planning. The lawsuits claim direct harm to health and safety and violations of consumer protection laws, which fall under harms to persons and communities. The AI system's design and use are central to the alleged harms. Despite the legal process being ongoing, the allegations describe realized harm and legal claims, meeting the criteria for an AI Incident rather than a hazard or complementary information. The event is not merely about potential harm or responses but about alleged actual harm caused by the AI system's use.[AI generated]

An AI system (ChatGPT) is explicitly involved as the patient used it to generate a medical diagnosis and demanded treatment based on it. The physician's account shows that the AI output was used in the patient's decision-making, but the physician did not follow the AI's advice blindly and insisted on clinical examination. No injury or harm has been reported, but the patient's overreliance on AI and refusal to allow proper medical assessment could plausibly lead to harm if acted upon. This fits the definition of an AI Hazard, where the AI system's use could plausibly lead to harm, but no incident has yet occurred. The event is not Complementary Information because it is not about a response or update to a prior incident, nor is it Beneficial Use since the AI is not used as a countermeasure but as a source of diagnosis by the patient. It is not Unrelated because AI is central to the event.[AI generated]

The article explicitly discusses AI systems designed for autonomous mission intelligence and coordination in military contexts, which clearly involve AI systems. Although no harm has yet occurred, the development and anticipated deployment of these systems in warfare plausibly could lead to significant harms such as injury, disruption, or violations of human rights. The article does not describe any actual incident or harm caused by these AI systems, so it is not an AI Incident. It is not merely complementary information because the main focus is on the development and market growth of potentially harmful AI defense technologies, not on responses or updates to past incidents. It is not unrelated or beneficial use, as the AI systems are intended for military offensive and defensive operations, which inherently carry risk of harm. Thus, the classification as an AI Hazard is appropriate.[AI generated]

The report explicitly states that AI was used in ransomware attacks affecting organizations, causing harm such as repeated attacks and ransom payments. The AI system's use in facilitating these attacks directly leads to harm to property and disruption of operations, fitting the definition of an AI Incident. The involvement of AI in the attack methods and the resulting realized harm confirms this classification.[AI generated]

The video was produced using AI (deepfake technology) to create false and misleading content that harms the reputation and honor of a political candidate, which constitutes a violation of rights and harms the legitimacy of the electoral process. This is a direct harm caused by the AI system's use. The court's decision to remove the video confirms the recognition of this harm. Hence, this qualifies as an AI Incident because the AI system's use directly led to harm (violation of rights and harm to communities).[AI generated]

The event involves the use of a drone (likely AI-enabled or autonomous) to attack a data center that supports AI development. The attack is a direct physical assault on critical infrastructure related to AI. Although the harm is caused by the attack itself, the AI system's development and use at the data center is central to the event. This qualifies as an AI Incident because the development and use of AI systems at the targeted data center is directly impacted by the attack, causing disruption to critical infrastructure. The presence of AI systems at the data center and the attack's impact on them meets the criteria for an AI Incident rather than a hazard or complementary information.[AI generated]

The event involves AI systems (text-to-image models) whose malfunction or inadequate safety guardrails have directly led to the generation and spread of harmful explicit content, including non-consensual intimate images and CSAM. This constitutes a violation of human rights and harm to communities, fulfilling the criteria for an AI Incident. The article documents realized harm, not just potential harm, and discusses the direct role of AI system vulnerabilities in enabling this harm. Therefore, the event is classified as an AI Incident.[AI generated]

The event involves an AI system explicitly used by the suspect to gather detailed information about company structures and meetings to plan espionage, which is a misuse of AI leading to potential harm to intellectual property and confidential business information. The installation of covert recording devices and the use of AI in this context directly or indirectly led to a breach of company security and potential harm to property and intellectual property rights. The company's use of AI to identify the suspect is a response but does not negate the incident classification. The harm is materialized or at least strongly implied, meeting the criteria for an AI Incident rather than a hazard or complementary information.[AI generated]

An AI agent is explicitly mentioned as having deleted the finished season, which is a malfunction of an AI system. This malfunction directly caused the loss of property (the edited season) and disruption to the media production schedule. According to the OECD framework, harm to property or communities includes harm to media content and economic assets. Therefore, this qualifies as an AI Incident due to direct harm caused by the AI system's malfunction. The harm is realized, not just potential, so it is not a hazard. It is not complementary information or unrelated, and it is not a beneficial use since the AI caused harm rather than prevented it.[AI generated]

The article explicitly mentions AI systems (self-replicating AI agents) being used maliciously in cyber fraud, which could plausibly lead to significant harm including financial losses and societal disruption. Since no specific incident of harm is detailed as having occurred, but credible warnings about potential harm are presented, this fits the definition of an AI Hazard. The article focuses on the plausible future risk of AI-enabled cyber fraud undermining trust in money and the financial system, rather than reporting a concrete AI Incident or a response to one.[AI generated]

An AI system was explicitly involved in generating the provocative image that triggered the racist harassment. The harm (racist abuse) is a direct injury to the player's well-being and dignity, fulfilling the harm to a person criterion. The AI system's output indirectly led to this harm by providing the content that sparked the attacks. The event is not merely a potential hazard or complementary information, as the harm has occurred. It is not unrelated or beneficial use, as the AI-generated content caused or contributed to the harm. Hence, the classification as AI Incident is appropriate.[AI generated]

The event involves AI systems and their development and use, with a focus on potential future security incidents that could arise from these systems. Since no actual harm has yet occurred but there is a credible and plausible risk of significant security incidents caused by AI in the near future, this fits the definition of an AI Hazard. The article is not reporting an AI Incident because the harms are prospective, not realized. It is not merely complementary information because the main focus is on the plausible future harm from AI systems, not on responses or updates to past incidents.[AI generated]

An AI system (Claude) was explicitly involved as the accused used it to obtain ideas for committing murder, which is a serious harm to a person. The event describes the use of the AI system in the development of a criminal plan, leading to legal action and arrest. This fits the definition of an AI Incident because the AI system's use indirectly led to a potential harm (planning of murder). The harm is materialized in the form of a criminal investigation and arrest, indicating the AI's role was pivotal in the incident. The article does not describe a mere potential risk or a societal response but an actual event involving harm and AI use.[AI generated]

An AI system was explicitly used to generate a fake image that caused real social harm by frightening people and triggering emergency responses. This fits the definition of an AI Incident because the AI system's use directly led to harm to the community (fear, disruption, and unnecessary mobilization of resources). Although the harm is non-physical, it is significant and clearly articulated. Therefore, the event qualifies as an AI Incident.[AI generated]

The AI system was explicitly used during the exam to cheat, which is a misuse of AI leading to a violation of legal frameworks protecting fair competition and rights in examination settings. This constitutes a breach of obligations under applicable law, fitting the definition of an AI Incident. The harm is realized as the integrity of the exam process is compromised, and the event involves direct use of an AI system causing this harm.[AI generated]

The article explicitly mentions AI systems developed for military applications that predict attacks and assist in planning artillery operations. These systems influence decisions that can cause injury, death, or property damage, fulfilling the criteria for harm (a) and (d). The AI systems are actively used or intended for use in combat, so their development and deployment are directly linked to harm. Hence, this qualifies as an AI Incident rather than a hazard or complementary information.[AI generated]

The event involves AI systems explicitly mentioned as machine-learning detection models used to identify fraudulent accounts on dating platforms. The failure or inadequacy of these AI systems to detect scammers directly led to significant financial harm to users, fulfilling the criteria for an AI Incident. The harm includes financial loss and emotional trauma, which are harms to persons and communities. The article also details ongoing improvements to the AI systems, but the primary focus is on the realized harm caused by the AI system's shortcomings, not just potential or future harm or complementary information about responses alone.[AI generated]

The article does not report any realized harm or incident caused by AI systems in utility companies but highlights concerns about possible risks such as cybersecurity breaches, data privacy violations, and algorithmic errors. The call for a review and regulatory oversight is a proactive measure to address potential future harms. Therefore, this event fits the definition of an AI Hazard, as it concerns plausible future harm stemming from AI use in critical infrastructure, without evidence of actual harm yet.[AI generated]

The article explicitly mentions terrorist groups exploiting AI, drones, and cryptocurrencies to enhance their operational capabilities, which constitutes the use of AI systems. However, it does not describe a specific AI Incident where harm has already occurred due to AI misuse. Instead, it raises alarms about the plausible future harms these technologies could cause if not properly countered. Therefore, this event fits the definition of an AI Hazard, as it plausibly could lead to AI-related harms such as security breaches, coordinated attacks, and evasion of law enforcement.[AI generated]

The CEO of Tving explicitly states that the hacking is presumed to be AI-based, indicating the involvement of an AI system in the attack. The data breach has already occurred, causing harm through the exposure of personal data, which is a violation of privacy rights and applicable law. This meets the criteria for an AI Incident because the AI system's use (in the hacking) has directly led to harm (violation of rights and breach of law). The investigation and potential sanctions are responses to this incident, not the primary event. The Google Korea data leak is unrelated to AI and thus does not affect the classification.[AI generated]