AIM: AI Incidents and Hazards Monitor

The article explicitly states that AI was used to develop a working zero-day exploit, which is a direct cause of a planned mass exploitation event. The exploit's AI-generated nature is confirmed by multiple indicators such as educational docstrings and AI-specific coding patterns. The event involves the use of AI in the development and intended deployment of a cyberattack tool, which constitutes harm to property and potentially critical infrastructure. The fact that the attack was thwarted does not negate the incident classification, as the AI system's use directly led to a significant cybersecurity threat. This fits the definition of an AI Incident because the AI system's use directly led to a harmful event (planned mass exploitation) that was narrowly avoided but clearly materialized as a serious threat.[AI generated]

Netflix's tracking and selling of user data, especially of children, implies the use of AI or algorithmic systems for data analysis and targeted advertising. The alleged deceptive practices and addictive design features (like autoplay) contribute to harm by exploiting children and violating privacy rights. This constitutes a violation of rights and harm to groups of people, meeting the criteria for an AI Incident. The lawsuit's focus on harm already caused and legal violations supports classification as an AI Incident rather than a hazard or complementary information.[AI generated]

TikTok's platform uses AI systems for personalized content recommendation and continuous scrolling, which are explicitly mentioned as mechanisms causing psychological harm to minors by exposing them to morbid and harmful content. The harm includes mental health deterioration and suicidal behavior, which are direct harms to individuals' health and well-being. The families' legal actions and ongoing investigations further confirm the realized harm linked to the AI system's use. Hence, this event meets the criteria for an AI Incident.[AI generated]

The article explicitly mentions the use of AI by students to cheat via covert devices communicating with AI systems. The frequency detectors are intended to prevent this misuse. Since no actual cheating incidents or harms have been reported yet, but the risk is credible and the measures are preventive, this qualifies as an AI Hazard. It is not an AI Incident because harm has not yet materialized, nor is it Complementary Information or Unrelated, as the focus is on a specific plausible AI-related risk and countermeasure.[AI generated]

The event involves an AI system (ChatGPT) whose use is alleged to have directly contributed to a mass shooting causing injury and death, which constitutes harm to persons. The lawsuit claims that the AI system acted as a co-conspirator by providing information used in planning the attack. Although OpenAI denies responsibility, the event meets the definition of an AI Incident because the AI system's use is linked to realized harm. Therefore, this is classified as an AI Incident.[AI generated]

The AI system Claude Mutos is explicitly mentioned as having capabilities to find vulnerabilities that could be exploited in cyberattacks, which could plausibly lead to disruption of critical infrastructure (financial systems). However, the article does not report any actual harm or incident occurring yet, only concerns and planned discussions to prevent such harm. Therefore, this qualifies as an AI Hazard, as the AI system's use or misuse could plausibly lead to an AI Incident involving disruption of critical infrastructure.[AI generated]

The event involves an AI system (Claude Mythos) designed to identify cybersecurity vulnerabilities, which could be used maliciously to cause harm. The article does not report any realized harm or incident but warns about the credible risk that AI-enabled cyberattacks could destabilize financial markets and cause significant economic harm. Therefore, this qualifies as an AI Hazard because it plausibly could lead to an AI Incident involving disruption of critical infrastructure and harm to financial stability, but no such incident has yet occurred.[AI generated]

The article explicitly mentions errors in AI services ChatGPT and Codex, which are AI systems. The malfunction is causing disruption in the use of these AI services, but there is no indication of harm to health, property, rights, or communities. The issue is a technical failure affecting service availability, which fits the definition of an AI Incident as it is a malfunction of AI systems causing disruption to their operation, even if the harm is limited to service disruption rather than physical or legal harm.[AI generated]

The article explicitly mentions AI systems deployed by Binance that have blocked $10.5 billion in crypto fraud, intercepted millions of scam attempts, and reduced fraud rates significantly. The AI systems' use in preventing realized financial harm to users and assisting in recovery and confiscation of illicit funds clearly indicates direct involvement of AI in mitigating an ongoing AI-related harm. Therefore, this qualifies as an AI Incident because the AI system's use is directly linked to preventing and addressing actual harm caused by AI-enabled fraudsters.[AI generated]

The article explicitly involves AI systems used in virtual companion apps that generate harmful and inappropriate content, including sexual and violent scenarios, some targeted or accessible to minors. The AI's outputs have directly led to harms such as exposure of minors to inappropriate content, emotional manipulation, and inducement to excessive paid consumption, which constitute violations of rights and harm to individuals and communities. The presence of AI-generated content with sexual and violent themes, the lack of effective age verification, and the inducement of minors to consume paid content demonstrate direct harm caused by the AI systems' use. Therefore, this qualifies as an AI Incident under the framework, as the AI system's use has directly led to significant harms including violations of rights and harm to health and communities.[AI generated]

The event involves the use of an AI system by a fiscal authority in decision-making processes that directly leads to harm, specifically violations of legal rights and procedural fairness for taxpayers. The AI-generated erroneous legal references and fabricated judicial decisions cause harm to individuals' rights and the justice process, fulfilling the criteria for an AI Incident under violations of human rights or breach of legal obligations. Therefore, this is classified as an AI Incident.[AI generated]

The article centers on the potential future risks posed by advanced AI models and the need for regulatory oversight to prevent such risks. It does not report any realized harm or incident involving AI systems but rather advocates for preventive safety reviews and enforcement mechanisms. Therefore, it fits the definition of an AI Hazard, as it concerns circumstances where AI development and use could plausibly lead to harm, specifically national security threats, if not properly managed.[AI generated]

The article details the creation of an AI system designed to predict fatigue through voice analysis, which is a novel application with potential safety benefits. Since no harm has occurred and the system is still in development/testing phases, this constitutes a plausible future risk scenario rather than an incident. Therefore, it fits the definition of an AI Hazard, as the AI system's use could plausibly lead to harm if it fails or is misused in critical safety contexts, but no direct or indirect harm is reported at this stage.[AI generated]

The article explicitly mentions the creation and distribution of sexually explicit AI deepfake videos exploiting Aman Gupta's persona, which constitutes a violation of his personality rights and trademarks. The AI system's use in generating deepfakes directly led to reputational harm and unauthorized commercial exploitation, fulfilling the criteria for an AI Incident. The court's injunction is a response to this realized harm, not merely a potential risk or complementary information.[AI generated]

The event involves the use of AI systems to generate false videos and content that are part of a disinformation campaign targeting a legislative proposal. The AI-generated misinformation has directly contributed to harm by misleading the public and fostering false narratives, which can be considered harm to communities and a violation of rights. Therefore, the event meets the criteria for an AI Incident due to the realized harm caused by AI-generated disinformation.[AI generated]

The event involves the use of AI systems to analyze AI-manipulated content that spreads false information during elections, which is a direct factor in combating harm to communities and the electoral process. The police are actively using AI to detect and respond to such harms, indicating realized harm from AI-generated disinformation. Therefore, this qualifies as an AI Incident because the AI system's use is directly linked to addressing harms caused by AI-generated false content affecting the election.[AI generated]

The article explicitly mentions AI systems developed by Palantir being used by military forces in active conflict zones, with direct involvement in target identification and attack operations that have caused harm and casualties. This meets the definition of an AI Incident because the AI system's use has directly led to injury and harm to groups of people, as well as violations of human rights. The article also discusses the AI systems' role in surveillance and lethal operations, confirming the AI system's pivotal role in causing harm. Hence, this is classified as an AI Incident.[AI generated]

The event involves AI systems explicitly generating hyper-realistic avatars that spread political messaging, which is a direct use of AI. The harm is realized as these AI-generated influencers are actively shaping public opinion and potentially distorting electoral outcomes, which qualifies as harm to communities and a violation of democratic rights. The article provides evidence of ongoing dissemination and influence, not just potential risk, thus meeting the criteria for an AI Incident rather than a hazard or complementary information. The involvement of AI in creating deceptive political influencers that manipulate public discourse is central to the harm described.[AI generated]

While the article references the development of artificial intelligence and new viralization techniques as challenges in combating electoral disinformation, it does not report a specific event where an AI system directly or indirectly caused harm or disruption. The concerns are about plausible future risks related to AI-enabled viralization methods and changes in governance that may affect oversight. Therefore, the event fits the definition of an AI Hazard, as it plausibly could lead to AI-related harms (e.g., disinformation impacting elections), but no realized harm or incident is described.[AI generated]

The event explicitly involves an AI system (deepfake technology) used to create a sexualized video of a person without consent, which is a violation of rights and causes harm to the individual and community. The harm has already occurred, as evidenced by the sharing of the video and subsequent disciplinary measures. This fits the definition of an AI Incident because the AI system's use directly led to a violation of rights and harm to a person.[AI generated]