AIM: AI Incidents and Hazards Monitor

The event involves AI systems (large language models powering chatbots) whose outputs reflect and extend government censorship, limiting political criticism of authoritarian leaders. This constitutes a violation of human rights, specifically freedom of expression, as the AI systems' behavior restricts speech beyond the jurisdictions where such restrictions legally apply. The harm is realized and ongoing, not merely potential, as the AI systems are actively producing censored outputs. Hence, this qualifies as an AI Incident due to indirect harm to human rights caused by the AI systems' use and design.[AI generated]

The article involves an AI system (TikTok's age inference technology) used to assess user age and enforce child safety measures. The investigation by Ofcom is prompted by doubts about the AI system's effectiveness, which could plausibly lead to harm (children accessing harmful content). Since no actual harm is reported but there is a credible risk of harm due to potential failure of the AI system, this event qualifies as an AI Hazard. It is not an AI Incident because no realized harm is described, nor is it Complementary Information since the main focus is the investigation and potential risk, not a response or update to a past incident.[AI generated]

The event clearly describes the use of AI applications to generate non-consensual explicit images, which constitutes a violation of the affected individuals' rights and moral integrity. The harm is realized and directly linked to the AI system's use in creating manipulated images. This fits the definition of an AI Incident due to violations of human rights and harm to individuals caused by the AI system's use.[AI generated]

TikTok uses AI systems for content recommendation and moderation, which are relevant to the protection of children from harmful content. The investigation concerns whether TikTok has fulfilled its legal duties, including effective age verification, which involves AI or algorithmic systems. Since no actual harm or violation has been confirmed yet, and the event centers on the potential failure to protect children, this fits the definition of an AI Hazard. It is not an AI Incident because harm has not been directly or indirectly established. It is not Complementary Information because the article is about the initiation of an investigation, not a response or update to a past incident. It is not Beneficial Use or Unrelated.[AI generated]

The event explicitly describes the use of an AI platform to edit a woman's photo into a nude image without consent, which was then posted online to harass and deceive. This is a clear case of AI misuse leading to violations of human rights and personal dignity, fitting the definition of an AI Incident. The harm is realized, not just potential, and the AI system's role is pivotal in generating the manipulated content. Therefore, this event qualifies as an AI Incident.[AI generated]

The article explicitly mentions the development of a massive AI data center supporting AI systems, which involves AI system use. The concerns raised by residents about water supply, power grid impact, and environmental preservation indicate plausible future harms linked to the AI infrastructure's operation. No actual harm or incident is reported; rather, the event centers on potential risks and community opposition. Hence, it qualifies as an AI Hazard rather than an AI Incident or Complementary Information. It is not unrelated or beneficial use, as the AI system is central to the event and the concerns are about potential harm.[AI generated]

The event involves the use of generative AI in producing legal citations that were fabricated or hallucinated, which directly led to harm in the form of misleading the court and potentially affecting the plaintiff's case. The judge explicitly raised concerns about AI use and the need for attorneys to verify AI-generated content to maintain candor and legal integrity. This constitutes a violation of legal obligations and harms the judicial process, meeting the criteria for an AI Incident. The harm is realized, not just potential, as the court had to address the issue and caution the government.[AI generated]

GPT-Red is an AI system explicitly described as being used internally to test and improve the security of other AI models by simulating attacks. While it has the capability to cause harm if misused, the article does not report any actual harm or incidents caused by GPT-Red. Instead, it highlights the potential risks and the precautions taken to avoid misuse. Therefore, this event represents a plausible future risk (hazard) due to the powerful offensive capabilities of GPT-Red and the concern about it falling into malicious hands. It is not an incident because no harm has occurred, nor is it complementary information since the main focus is on the AI system's development and its potential risks rather than a response or update to a past incident. It is not unrelated or beneficial use, as the AI system itself is designed for offensive security testing, which could plausibly lead to harm if misused.[AI generated]

The article centers on a claim that AI might have been used to fabricate images related to a tragic event, implying a potential for AI-generated misinformation. There is no evidence that such AI-generated images have caused harm yet, only a suggestion that this could be the case. The AI system's involvement is speculative and concerns potential misuse leading to harm (disinformation about a mass casualty event). This aligns with the definition of an AI Hazard, where AI use could plausibly lead to harm but no harm has been confirmed or directly linked to AI use. Hence, the event is classified as an AI Hazard rather than an AI Incident or Complementary Information.[AI generated]

The event involves AI systems (driver-assistance and self-driving technologies) whose misuse or misunderstanding by humans has directly contributed to road safety risks and at least one fatal accident. This constitutes harm to persons (injury or death), fulfilling the criteria for an AI Incident. The article reports on realized harms and ongoing concerns, not just potential risks, so it is not merely an AI Hazard or Complementary Information. Therefore, the classification is AI Incident.[AI generated]

The article explicitly states that AI-powered drones are responsible for killing Russian soldiers on the battlefield, with an average life expectancy of 20 to 30 minutes for recruits due to these drones. This is a direct harm to human life caused by the use of AI systems in military operations, fitting the definition of an AI Incident under harm to health and injury to groups of people. The involvement is clear and direct, and the harm is realized, not just potential.[AI generated]

The event involves AI systems used by fraudsters to conduct phishing scams, which directly cause harm to individuals by stealing sensitive information and money, fulfilling the criteria for harm to persons (a). The AI's role is pivotal as it enables the scale and personalization of these scams. The article documents actual harm occurring, not just potential risks or responses, so it is an AI Incident rather than a hazard or complementary information. The mention of AI-based security systems is background context and does not negate the presence of an AI Incident caused by malicious AI use.[AI generated]

The article centers on the potential future harms that could plausibly arise from the use or misuse of advanced AI systems like Mythos, particularly in enabling cyberattacks on critical infrastructure. There is no indication that an actual incident has occurred yet, but the concerns expressed by a major industry leader and the US government's prior restrictions demonstrate a credible risk. The AI system's development and use could plausibly lead to an AI Incident involving disruption of critical infrastructure. Therefore, this event qualifies as an AI Hazard rather than an AI Incident or other category.[AI generated]

The article involves an AI system developed by Google DeepMind and its use in a Pentagon contract for classified AI applications. The employee's resignation and warnings highlight concerns about potential misuse of AI for harmful military purposes, including mass surveillance and autonomous weapons, which could violate human rights or cause harm to communities. Since the contract has been signed but no specific harm has been reported as realized, the event fits the definition of an AI Hazard rather than an AI Incident. The employee's resignation and petition by other employees further underscore the plausible risk of harm from this AI use.[AI generated]

The event involves an AI system in the form of a data center intended to support AI workloads, which could plausibly lead to environmental and community harms if built and operated as planned. However, since the data center is still in the proposal and approval stage with no actual harm reported, this fits the definition of an AI Hazard rather than an AI Incident. The article does not focus on a societal or governance response alone, so it is not Complementary Information. It is not unrelated or a beneficial use, as the AI system is central to the discussion and the concerns are about potential harm.[AI generated]

The event involves an AI system (an LLM) explicitly used in the development of a malicious IoT botnet framework. The botnet is operational and actively infecting devices, enabling harmful activities such as DDoS attacks and unauthorized access, which constitute harm to property and communities. The AI system's involvement in generating code that forms the botnet framework is a direct contributing factor to the harm caused. Although some bugs exist due to AI-generated code, the botnet is functional and deployed, with evidence of ongoing infections and active C2 infrastructure. This meets the criteria for an AI Incident, as the AI system's use has directly led to significant harm. The detailed analysis and mitigation advice do not change the classification, as the primary event is the harmful deployment of an AI-assisted malware framework.[AI generated]

The event involves AI biometric identification systems used by government agencies, suspected to be Chinese-made or rebranded Chinese products, which may violate procurement rules and pose national security risks. The AI system's use here could plausibly lead to harm through data leakage of unique biometric identifiers, which cannot be reset, thus constituting a serious privacy and security risk. Since no actual data breach or harm is reported yet, but the risk is credible and significant, the event fits the definition of an AI Hazard rather than an AI Incident. It is not Complementary Information because the article focuses on revealing the potential risk, not on responses or updates to a past incident. It is not Beneficial Use or Unrelated.[AI generated]

The article explicitly describes AI systems used in medical diagnostics and knowledge work that, through their use, have caused a decline in human professional skills and critical thinking. This decline has been measured and documented, showing a direct link between AI use and harm to human capabilities essential for health and safety. The harm is indirect but clearly articulated and pivotal, fitting the definition of an AI Incident. The article does not merely warn of potential harm (which would be an AI Hazard) nor is it a societal or governance response (Complementary Information). It is not unrelated or solely beneficial use, as the AI's role leads to a significant negative impact on human skills and potentially on health outcomes.[AI generated]

The article explicitly involves an AI system (Google's AI-powered search engine and chatbot) whose use has directly led to harm to children, including failure to detect and respond appropriately to suicidal ideation and eating disorders, validation of harmful behaviors, and dissemination of harmful instructions. These harms fall under injury or harm to health and harm to communities. The AI system's malfunction or inadequate safeguards are central to these harms. Although Google disputes some findings, the investigation's detailed examples and testing indicate realized harm. Hence, this event meets the criteria for an AI Incident rather than a hazard or complementary information.[AI generated]

The event explicitly involves an AI system (Grok AI chatbot) being misused to generate harmful content, including sexual deepfakes and CSAM, which are serious violations of rights and cause emotional harm to victims. The AI system's involvement is central, as it was used to create the illicit content, and the harms have materialized, including legal actions and emotional distress. The misuse and resulting harm fit the definition of an AI Incident, as the AI system's use directly led to violations of human rights and harm to individuals and communities.[AI generated]