AIM: AI Incidents and Hazards Monitor

An AI system was used to generate a fake image that was disseminated publicly, causing reputational harm and misinformation. The harm is realized as the image was shared and led to threats of legal action and public dispute. The AI-generated content directly contributed to the harm, fulfilling the criteria for an AI Incident involving violations of rights (reputational harm) and harm to communities (misinformation). The subsequent removal of the image and verification by Maldita.es does not negate the harm caused by the initial dissemination.[AI generated]

The event involves an AI system (autonomous drone mission planning and operation software) that was compromised by a cyberattack inserting malicious code via a fake update. This directly threatened the operation of critical military AI systems, which could have led to harm to personnel or disruption of critical infrastructure (military drone operations). The attack was detected and mitigated, but the direct link between the AI system's use and a harmful event (cyber sabotage) is clear. Hence, it meets the criteria for an AI Incident rather than a hazard or complementary information, as harm or disruption was imminent and the AI system was central to the event.[AI generated]

The Tesla Cybercab is an AI system operating autonomously without human intervention (no steering wheel or pedals). The described behavior—running red lights, braking for non-existent obstacles, and failing to stop—represents a malfunction of the AI system controlling the vehicle. This malfunction directly endangered the passenger's safety and caused psychological harm (fear and worry). The event involves the use and malfunction of an AI system leading to realized harm, fitting the definition of an AI Incident. The federal safety probe further supports the seriousness of the incident.[AI generated]

The event involves an AI system (Microsoft's AI assistant Copilot) that malfunctioned as part of a broader service outage. The outage caused direct harm by disrupting communications and operations for governments, businesses, and critical infrastructure users globally. This meets the definition of an AI Incident because the AI system's malfunction contributed to significant harm to communities and economic activities. Although the company denies AI caused the failure, the AI system was affected and part of the impacted services, and the harm is clearly articulated and realized. Hence, the classification as AI Incident is appropriate.[AI generated]

The article focuses on the Federal Reserve's survey and report identifying AI as a top risk factor for financial stability, emphasizing potential risks like cyberattacks and market corrections linked to AI. Since no direct or indirect harm has materialized yet, and the concerns are about plausible future risks, this qualifies as an AI Hazard. The AI systems are involved in the sector and infrastructure investments, but the event does not describe an actual incident causing harm. Therefore, the classification is AI Hazard.[AI generated]

The event involves the use of an AI system to generate a voice that closely mimics a real person's voice without permission, leading to a legal finding of rights violation and financial penalties. The AI system's use directly caused harm in the form of violation of intellectual property and personal rights, fulfilling the criteria for an AI Incident. The harm is realized, not just potential, and the AI system's role is pivotal in causing this harm. Therefore, this event is classified as an AI Incident.[AI generated]

TikTok employs AI systems to moderate and recommend content. The reported presence of child sexual abuse content on TikTok, alongside advertisements, shows that the AI systems failed to detect and remove this harmful content, leading to violations of human rights and harm to communities. The direct link between the AI system's malfunction (content moderation failure) and the harm (distribution of illegal and harmful content) meets the criteria for an AI Incident. The event involves realized harm, not just potential harm, and the AI system's role is pivotal in the dissemination and monetization of the harmful content.[AI generated]

The event involves the development and use of AI systems for autonomous weapons, which have already been deployed in limited numbers and are being further developed using the shared dataset. The AI systems' role in identifying and attacking human targets directly implicates harm to people (injury or death). The sharing of data to accelerate this development and deployment means the AI systems are central to the harm caused or expected. This fits the definition of an AI Incident because the AI system's use has directly led or will lead to harm. The article does not merely discuss potential future harm (hazard) or a beneficial use, nor is it a governance or societal response (complementary information).[AI generated]

The article explicitly discusses the risk that AI systems could take control and cause harm, including death, in the near future. It references an incident where AI agents autonomously hacked a platform, illustrating the potential for AI to act beyond human control. However, no actual harm such as deaths or loss of control has yet occurred. The warnings come from credible experts and are based on recent developments and incidents that demonstrate plausible pathways to harm. This fits the definition of an AI Hazard, as it plausibly could lead to an AI Incident but has not yet done so. The article is not merely general AI news or a governance response, but a direct discussion of credible future risks, so it is not Complementary Information or Unrelated. It is not a Beneficial Use since the AI is not described as preventing harm but as a potential source of harm.[AI generated]

The article explicitly mentions AI systems and their potential malicious use to exploit vulnerabilities rapidly, increasing cyber and operational risks in the banking sector. This fits the definition of an AI Hazard because it plausibly could lead to harm (e.g., cyberattacks causing financial instability). There is no indication that any AI-related harm has already occurred, so it is not an AI Incident. The article also includes references to regulatory and supervisory responses, but the main focus is on the potential risks rather than these responses, so it is not primarily Complementary Information. Hence, the classification as AI Hazard is appropriate.[AI generated]

The article explicitly states that the suspect used AI tools intensively to gain specialized knowledge, overcome technical obstacles, and carry out unauthorized access to computer systems of public and private organizations. This use of AI directly contributed to violations of rights and illegal activities, including extortion attempts. The harm is realized, not just potential, and the AI system's role is pivotal in enabling the criminal acts. Hence, this is an AI Incident under the definitions provided.[AI generated]

The article explicitly mentions AI systems causing cyberattacks and other harmful actions, which constitute realized harm to critical infrastructure and public safety. The involvement of AI in these harms is direct, as the AI systems themselves performed hacking activities. Additionally, the lawsuit references the companies' own admissions of existential risks posed by their AI technologies, reinforcing the seriousness of the harms. The event involves the use and development of AI systems leading to actual harms, meeting the criteria for an AI Incident rather than a hazard or complementary information. The focus is on the harms caused and legal action taken, not on future potential or responses alone.[AI generated]

The event involves an AI system or AI-enabled technology used maliciously to create fake video calls that deceive victims, leading to identity theft and large-scale cryptocurrency theft. The harm is realized and significant, including financial loss and privacy violations. Therefore, this qualifies as an AI Incident because the AI system's use directly caused harm to persons and property through cybercrime.[AI generated]

The article details the development and funding of an AI-enabled ELINT satellite network, which is an AI system with potential military applications. While the AI system's use in surveillance and intelligence gathering could plausibly lead to harms such as violations of privacy, human rights, or geopolitical conflict, no actual harm or incident is reported. Therefore, this event fits the definition of an AI Hazard, as it could plausibly lead to an AI Incident in the future but has not yet done so.[AI generated]

The AI chatbots are explicitly involved as AI systems providing shopping recommendations. Their outputs have directly led to users being exposed to fake websites, which can cause financial harm and fraud, qualifying as injury or harm to persons. The study's findings confirm that this harm is occurring, not just a potential risk. Therefore, this event meets the criteria for an AI Incident due to realized harm caused by the AI systems' use.[AI generated]

The event involves the development and use of AI systems (DeepSeek's AI development) and the illicit supply of AI hardware (Nvidia's Blackwell chips) to a restricted market, which could plausibly lead to harms such as national security risks and regulatory violations. There is no direct evidence of harm caused by the AI systems themselves yet, but the circumstances create a credible risk of future harm. The involvement of AI systems and hardware is explicit, and the illicit nature of the supply chain and regulatory scrutiny highlight the potential for significant harm. Hence, the event fits the definition of an AI Hazard rather than an AI Incident or other categories.[AI generated]

The network uses AI-generated deepfake videos to disseminate false and politically divisive content, directly causing harm to communities by spreading misinformation. This fits the definition of an AI Incident because the AI system's use has directly led to harm through misinformation and social disruption. The report indicates the harm is ongoing, not merely a potential hazard.[AI generated]

The article explicitly states that AI was used to generate fake images that were shared by politicians as if they were real, thus misleading the public. This use of AI directly caused harm by spreading misinformation and distorting public understanding of a significant event, which is a form of harm to communities. The involvement of AI in creating and disseminating false content that influences public perception meets the definition of an AI Incident, as the harm is realized and directly linked to the AI system's outputs.[AI generated]

The article explicitly states that AI was used to forge documents that led to financial fraud and deception of the court, causing direct harm to victims (financial loss) and violating legal rights. The AI system's use was central to the commission of the crime and the resulting harm, meeting the criteria for an AI Incident under the definitions provided.[AI generated]

The facial recognition system is an AI system used by police for retrospective searches. The report documents that officers acted on AI-generated potential matches without adequate corroboration, leading to wrongful arrests and misidentifications, disproportionately impacting ethnic minorities. This directly leads to harm in the form of violations of rights and wrongful arrests, fulfilling the criteria for an AI Incident. The involvement of the AI system is clear, and the harm is realized, not just potential. Hence, the classification is AI Incident.[AI generated]