AIM: AI Incidents and Hazards Monitor

Tesla’s FSD is an AI-driven driver-assist system whose malfunction (failure to register an SUV in a blind spot) directly contributed to a collision with minor injuries and a totaled car. The incident demonstrates realized harm from an AI system in use, making this an AI Incident.[AI generated]

The event involves AI systems explicitly, specifically AI tools used to generate harmful sexual imagery and deepfakes. The misuse of these AI systems has directly led to harms including violations of human rights and sexual exploitation of minors and adults, which are clear harms under the AI Incident definition. The laws are a response to realized harms caused by AI-generated content, not just potential future harms. Therefore, this event qualifies as an AI Incident because it concerns the direct use of AI systems to produce illegal and harmful content affecting individuals' rights and safety.[AI generated]

YouTube's recommendation system is an AI system that influences user engagement by steering users towards certain content. The lawsuit claims this system amplifies harmful material and drives addictive behavior among youth, leading to mental health harms and exposure to harmful content. These harms fall under injury or harm to health and harm to communities. The AI system's use is directly linked to these harms, fulfilling the criteria for an AI Incident. The event is not merely a potential risk or a governance response but describes ongoing harm caused by the AI system's operation.[AI generated]

The KARGU drone is an AI system as it uses AI and image processing capabilities for autonomous navigation and target engagement. Its use in combat with a new armor-piercing warhead that successfully destroyed a target demonstrates direct harm caused by the AI system's operation. The article explicitly states the system's operational use in military and counter-terrorism contexts, implying realized harm to property and potentially persons. Hence, this is an AI Incident rather than a hazard or complementary information.[AI generated]

While the study demonstrates that AI can be used to recreate screen content (including sensitive credentials) from HDMI emissions, there is no specific, confirmed incident of harm using this exact method. Rather, it outlines a credible future threat—i.e. a novel AI-enabled espionage attack—making it a potential hazard rather than a realized incident or a mere background update.[AI generated]

The article involves AI systems implicitly through the technology in electric vehicles and solar inverters that include software capable of remote updates, tracking, and control, which can be reasonably inferred to involve AI or advanced algorithmic systems. The concerns raised relate to the potential misuse or malicious use of these AI-enabled systems to cause harm such as disruption or privacy violations. However, no actual incident or harm has occurred; the fears are about plausible future misuse. Therefore, this qualifies as an AI Hazard, as the development and use of these AI systems could plausibly lead to an AI Incident involving national security or privacy harms, but no direct or indirect harm has yet materialized.[AI generated]

The event involves a deployed AI system (a Waymo driverless car) whose use directly led to harm (vandalism, emotional trauma, risk to passengers). Although the damage was inflicted by humans, it targeted the AI-enabled vehicle and threatened passenger safety, qualifying as an AI Incident under harm to persons and property.[AI generated]

The article describes a realized harm caused by the misuse of AI tools: two minors used AI to create and post a fake obscene image of their teacher, infringing her rights and causing reputational and psychological harm. This constitutes a direct AI Incident under the framework.[AI generated]

The article explicitly mentions the development and planned validation of an autonomous AI system for mining trucks, which qualifies as an AI system. However, there is no mention of any injury, disruption, rights violation, or other harm caused by the system so far. The event concerns the development and prospective use of the AI system, which could plausibly lead to future harms (e.g., safety risks in autonomous mining vehicles), but no incident or harm has yet occurred. Therefore, this event fits the definition of an AI Hazard, as it plausibly could lead to harm in the future but does not describe an actual incident or realized harm.[AI generated]

The article explicitly mentions AI-generated photos being used to spread false narratives and manipulate social media users, which is a direct use of AI systems (generative AI) leading to harm in the form of misinformation and potential fraud. This fits the definition of an AI Incident as the AI system's use has directly led to harm to communities and potential violations of rights through manipulation and fraud. The involvement of bots spreading these posts further supports the AI system's role in causing harm.[AI generated]

The article describes the actual use of an AI system to fabricate WhatsApp conversations, falsely accuse and threaten a victim, and extort funds. This misuse directly caused psychological harm, harassment, and fraud, meeting the definition of an AI Incident.[AI generated]

The incident involved active misuse of AI (an AI-generated deepfake avatar) to perpetrate a cryptocurrency scam, directly causing potential financial harm to viewers. Since an AI system’s use directly led to real-world harm, this qualifies as an AI Incident.[AI generated]

The event describes an actual malicious use of AI-driven deepfake technology to deceive a senator’s office, posing national security and political risks. This is a direct harm scenario where AI-generated content was used to influence and extract sensitive information, fitting the definition of an AI Incident.[AI generated]

The article explicitly discusses AI technologies and their potential to create new vulnerabilities in the financial system, which could plausibly lead to harm such as financial instability or systemic risk. No actual harm or incident is reported; the focus is on warnings and the need for policy adaptation. This fits the definition of an AI Hazard, as it concerns plausible future harm from AI use in finance. It is not Complementary Information because it is not updating or responding to a past incident, nor is it unrelated as it clearly involves AI systems and their impact.[AI generated]

The event involves AI systems indirectly through the demand for AI hardware but does not describe any realized harm or incident caused by AI system development, use, or malfunction. The report warns of a plausible future supply shortage due to AI-driven demand, which could lead to economic or operational disruptions if it occurs. However, since no actual harm or incident has yet occurred, and the article primarily provides an analysis and warning about potential future risks, it fits the definition of an AI Hazard rather than an AI Incident or Complementary Information.[AI generated]

The misuse of AI-generated voice cloning for robocalls to mislead and suppress voters constitutes a clear case where an AI system’s use directly led to harm (voter suppression, violation of election integrity). This is not merely a warning or update, but an actual incident of AI-caused harm.[AI generated]

The event involves an AI system (Google Gemini for Workspace) that is susceptible to indirect prompt injection attacks, which manipulate the AI's outputs in ways that could mislead users or facilitate phishing attacks. This constitutes a direct or indirect cause of harm to users' security and privacy, fitting the definition of an AI Incident due to violations of user rights and potential harm to individuals. The fact that the vulnerability has been exploited in proofs-of-concept and reported to Google, with no remediation, confirms the realized risk rather than a mere potential hazard.[AI generated]

The article explicitly mentions targeted algorithms used to promote alcohol content on social media, which can be reasonably inferred to involve AI systems for content recommendation and targeting. The exposure of young people to such advertising is linked to increased desire to consume alcohol, which is a health harm and social harm. The AI system's role in enabling this targeted exposure is pivotal in the chain of harm. Thus, this qualifies as an AI Incident due to indirect harm caused by AI-enabled targeted advertising promoting alcohol to minors.[AI generated]