Whistleblowers Expose Meta and TikTok's AI Algorithms Amplifying Harmful Content

The article explicitly discusses AI systems (algorithms for content recommendation and feed ranking) developed and used by Meta and TikTok. These AI systems have directly led to harms including psychological injury to users (especially adolescents), exposure to harmful and extremist content, and violations of user safety and rights. The harms are realized and documented through internal research and whistleblower accounts. The AI systems' prioritization of engagement over safety is a direct cause of these harms. Therefore, this event qualifies as an AI Incident because the development and use of AI systems have directly led to significant harm to individuals and communities.

The event explicitly involves AI systems in the form of recommendation algorithms used by TikTok and Meta to curate and promote content. The whistleblowers reveal that these AI systems were deliberately tuned or allowed to promote harmful content to increase user engagement, leading to real harm such as exposure to violent and hateful content and user radicalization. This meets the criteria for an AI Incident because the AI systems' use has directly or indirectly led to harm to individuals and communities, including mental health harm and violations of user safety rights. The presence of internal documents and user testimonies further supports the direct link between AI system use and harm.

The article explicitly discusses AI systems in the form of recommendation algorithms that influence content feeds on social media platforms. These AI systems were used in ways that knowingly or negligently amplified harmful content, causing direct harm to users, including psychological harm and exposure to dangerous material. The whistleblower evidence and internal documents confirm that the AI systems' outputs were a pivotal factor in these harms. Therefore, this event meets the definition of an AI Incident due to direct harm caused by the use of AI systems in content recommendation and moderation failures.

The event explicitly involves AI systems in the form of recommendation algorithms that select and promote content to users. The companies' deliberate decisions to allow harmful content to be amplified by these AI systems directly caused harm to users, including psychological harm and exposure to violent and hateful content. The harm is realized and documented through whistleblower testimonies and internal investigations. This meets the criteria for an AI Incident because the AI system's use led directly to violations of user safety and harm to communities.

The event explicitly involves AI systems in the form of recommendation algorithms that influence content feeds on social media platforms. The harms described include psychological harm, exposure to harmful and radicalizing content, and violations of user safety, especially for minors, which fall under harm to communities and individuals. The companies' internal knowledge and decisions to allow harmful content for engagement gains indicate direct and indirect causation of harm by the AI systems. This meets the criteria for an AI Incident rather than a hazard or complementary information, as the harms are ongoing and documented through whistleblower evidence and internal research.

The article explicitly involves AI systems in the form of recommendation algorithms and AI-based moderation tools used by Meta and TikTok. These AI systems have been used in ways that have directly or indirectly led to harm to communities and individuals, including exposure to harmful content, radicalization, and failure to protect minors. The harms are realized and documented through internal studies and whistleblower testimonies. The companies' decisions to prioritize engagement and stock prices over safety demonstrate the AI systems' role in causing these harms. Hence, the event meets the criteria for an AI Incident.

The article explicitly involves AI systems—recommendation algorithms used by Meta and TikTok—that have been used in ways that directly or indirectly led to harm to communities and individuals by promoting harmful content. The whistleblower testimonies and internal documents indicate that these harms are ongoing and were knowingly tolerated or insufficiently mitigated by the companies. This fits the definition of an AI Incident because the AI systems' use has directly or indirectly caused harm (harmful content exposure, radicalization, hate speech proliferation). The event is not merely a potential risk or a governance response but reports realized harms linked to AI system use.

The event explicitly involves AI systems (content recommendation algorithms) whose use directly led to harm to minors and communities through exposure to harmful content and radicalization. The harm is realized and ongoing, not merely potential. Therefore, this qualifies as an AI Incident under the framework, as the AI systems' use has directly led to harm to groups of people (minors) and communities through the spread of harmful content and radicalization.

The article explicitly mentions algorithmic strategies, which reasonably infer the involvement of AI systems in content recommendation and moderation. The harmful content spread includes violence, sexual harassment, terrorism, and divisive political content, which constitute harm to communities and potentially violations of rights. The AI systems' use to prioritize engagement over safety directly leads to these harms. Therefore, this event qualifies as an AI Incident due to the realized harm caused by the AI systems' deployment and management decisions.

The event explicitly involves AI systems in the form of content recommendation algorithms that influence what users see. The companies' decisions to allow more harmful content to increase engagement directly led to harm to users and communities, fulfilling the criteria for an AI Incident. The harms include exposure to harmful content such as violence, sexual blackmail, misogyny, conspiracy theories, bullying, and hate speech, which affect the health and well-being of individuals and communities. The involvement of AI in these harms is direct, as the algorithms determine content visibility and engagement. Therefore, this event qualifies as an AI Incident.

The event explicitly involves AI systems in the form of recommendation algorithms that influence content feeds on social media platforms. The harms described include injury to mental health, radicalization, exposure to harmful and violent content, and normalization of hate speech, which affect individuals and communities. The AI systems' development and use decisions, such as allowing more harmful content to increase engagement, directly contributed to these harms. The whistleblower evidence and internal research confirm the causal link between the AI algorithms and the harms. Hence, this is an AI Incident as the AI systems' use has directly and indirectly led to significant harms.

The article explicitly discusses AI-powered recommendation algorithms at TikTok and Meta that have been used to maximize user engagement by promoting 'borderline harmful content' and harmful posts. The whistleblowers provide evidence that these AI systems have directly contributed to the spread of harmful content, including sexual blackmail, cyberbullying, and hate speech, causing real harm to individuals and communities. The AI systems' development and use have led to violations of rights and harm to users, meeting the criteria for an AI Incident. The involvement of AI is clear from the description of recommendation engines and algorithmic content promotion. The harms are realized and ongoing, not merely potential.

The platforms' AI recommendation algorithms are AI systems that infer from user data to generate content feeds. The whistleblowers reveal that management decisions led to the AI systems being tuned to promote more harmful content, which directly caused harm to users and communities by exposing them to violence, sexual blackmail, and extremist content. This constitutes an AI Incident because the AI systems' use directly led to violations of safety and harm to communities. The event involves realized harm, not just potential risk, and thus is not merely a hazard or complementary information.

The article explicitly involves AI systems in the form of social media recommendation algorithms that select and promote content to users. The harm includes psychological injury, radicalization, normalization of violence, and exposure to harmful content, which affect individuals and communities, fulfilling the harm criteria. The AI systems' use and management decisions (e.g., instructions to allow harmful content) have directly led to these harms. Although companies deny or claim mitigation efforts, the reported harms are ongoing and documented. Hence, this is an AI Incident rather than a hazard or complementary information.

The event explicitly involves AI systems in the form of recommendation algorithms that select and promote content on social media platforms. The whistleblowers provide evidence that these AI systems were deliberately allowed or designed to promote harmful content to maximize engagement and profits, despite known risks. The harms include psychological harm to individuals (e.g., radicalization, exposure to sexual blackmail), harm to communities (normalization of hate speech, violence), and violations of user safety rights. The AI systems' outputs directly contributed to these harms, fulfilling the criteria for an AI Incident. The event is not merely a potential risk or a governance response but documents realized harms caused by AI system use and management decisions.

Instagram's Reels uses AI algorithms to recommend and display content. The internal research showing increased harmful content in Reels indicates that the AI system's outputs contributed to harm to communities and individuals (bullying, harassment, hate speech). The company's decision to prioritize engagement growth despite these findings and insufficient safety staffing implies a failure to mitigate known harms. Therefore, the AI system's use directly led to realized harm, fitting the definition of an AI Incident.

The social media platforms use AI systems (algorithms) to curate and prioritize content in users' feeds. The whistleblower reports indicate that these AI systems were deliberately tuned or allowed to promote harmful content to increase engagement, which has led to real harm such as exposure to misogynistic content, conspiracy theories, and neglect of reports of sexual exploitation. This meets the definition of an AI Incident as the AI system's use has directly or indirectly led to harm to communities and user safety. The involvement is in the use of AI systems for content recommendation and moderation decisions that resulted in harm.

The article explicitly discusses AI-powered recommendation algorithms at Meta and TikTok that optimize for engagement, leading to increased exposure to harmful content such as bullying, hate speech, and radicalizing material. The whistleblower accounts and internal research confirm that these harms are occurring and are a direct consequence of the AI systems' operation and the companies' decisions to loosen safety standards. The harms to individuals' mental health and to communities through the spread of extremist content fall under harm to health and harm to communities. Therefore, this event meets the criteria for an AI Incident due to the direct and indirect harms caused by the AI systems' use and malfunction (inability to control outputs).

The article explicitly discusses AI-powered recommendation systems and algorithms used by TikTok and Meta that promote harmful content to maximize engagement and profits, despite internal knowledge of the risks. The whistleblower accounts indicate that these AI systems' outputs have directly led to harm through the spread of harmful and dangerous content, including hate speech and harassment, which harms communities and individuals. This meets the criteria for an AI Incident as the AI systems' use has directly led to harm. The event is not merely a potential hazard or complementary information but a report of realized harm linked to AI system use.

The article explicitly discusses AI-powered recommendation algorithms used by Meta and TikTok that have been shown internally to promote harmful content to maximize engagement and profits, despite known risks to user safety. The whistleblowers provide evidence that these AI systems' outputs have directly led to harm by exposing users to harmful and dangerous content. The involvement of AI systems in causing harm to users and communities through content amplification fits the definition of an AI Incident, as the harm is realized and directly linked to the AI systems' use and management decisions.

The article explicitly discusses AI-powered recommendation algorithms at Meta and TikTok that amplified harmful content, causing direct harm to users, including exposure to violence, harassment, and radicalization. The whistleblowers describe internal decisions to prioritize engagement metrics over safety, leading to real and ongoing harm to individuals and communities. The involvement of AI systems in content recommendation and moderation is clear, and the harms described fall under injury or harm to persons and harm to communities. Therefore, this qualifies as an AI Incident rather than a hazard or complementary information.

The article explicitly discusses AI systems (machine-learning recommendation algorithms) used by Meta and TikTok that have been manipulated to prioritize engagement over safety, resulting in the amplification of harmful content. This has directly led to harms such as radicalization, normalization of hate speech, and exposure of users, including children and teenagers, to harmful content. The AI systems' role is pivotal in causing these harms by shaping what content users see. The harms include violations of community safety and potentially human rights related to exposure to harmful content. Hence, this is an AI Incident rather than a hazard or complementary information.

The event explicitly involves AI systems in the form of social media content recommendation algorithms that influence what content users see. The use and tuning of these AI systems directly led to the dissemination of harmful content, causing harm to users' wellbeing and communities. The whistleblower accounts and internal studies confirm that the AI systems' development and use prioritized profit over safety, resulting in real harm. Therefore, this qualifies as an AI Incident due to direct harm caused by the AI systems' outputs and their role in amplifying harmful content.

The article explicitly mentions AI-powered recommendation systems and algorithms that promoted harmful content to maximize engagement, leading to real harms such as mental health issues among teenagers, radicalization, and exposure to extremist content. The involvement of AI in content curation and amplification is central to the harm described. The harms include injury to health (mental health harms), harm to communities (spread of extremist and harmful content), and violations of user safety. The presence of lawsuits and whistleblower evidence confirms that harm has occurred, not just potential harm. Hence, this event meets the criteria for an AI Incident.

The platforms' AI recommendation algorithms, which infer user preferences and optimize content delivery, were intentionally tuned to maximize engagement at the expense of safety, directly causing harm to users, especially vulnerable groups like minors. The documented increase in bullying, harassment, hate speech, and exposure to extremist content constitutes harm to communities and individuals' health. The whistleblower evidence and internal research confirm that these harms are a direct consequence of AI system use and corporate policy decisions. Therefore, this qualifies as an AI Incident because the AI systems' use has directly led to significant harms as defined in the framework.

The whistleblower revelations indicate that AI systems used by Meta and TikTok for content recommendation and moderation are directly linked to the amplification of harmful content, causing harm to communities through increased exposure to violence, exploitation, extremism, and hate speech. This meets the criteria for an AI Incident as the AI systems' use has directly led to violations of rights and harm to communities. The other topics in the article do not involve AI systems or related harms and are therefore unrelated.

The article explicitly implicates AI-driven content recommendation systems used by Meta and TikTok in promoting harmful content that causes real harm to communities, including the spread of extremism and dangerous misinformation. The involvement of AI systems in content moderation and feed curation is reasonably inferred, and the resulting harm to communities is direct and ongoing. Therefore, this event qualifies as an AI Incident due to the direct link between AI system use and realized harm.

The event involves AI systems explicitly described as recommendation algorithms that influence content exposure on social media platforms. The use of these AI systems has directly led to harms including radicalization, exposure to hate speech, and psychological harm to users, fulfilling the criteria for harm to communities and individuals. The article provides evidence from whistleblowers and internal research showing that the AI systems prioritize engagement over user well-being, causing significant societal harm. Hence, it meets the definition of an AI Incident rather than a hazard or complementary information.

The article explicitly involves AI systems in the form of recommendation algorithms that influence content exposure. The harm includes exposure to harmful content, radicalization, and insufficient safety measures, which are direct consequences of the AI systems' operation and prioritization. These harms affect user health and safety and community well-being, fitting the definition of an AI Incident. The whistleblower reports and internal documents confirm the AI systems' role in causing these harms, not just potential future risks.

The article explicitly discusses AI-powered recommendation algorithms that influence the content users see, which is a clear AI system. The harm described includes exposure to harmful content, radicalization, cyberbullying, and normalization of hate and violence, which constitute harm to health, communities, and violations of rights. The platforms' decisions to allow or amplify such content for engagement and profit show the AI systems' use directly leading to these harms. Therefore, this qualifies as an AI Incident rather than a hazard or complementary information.

Social media platforms employ AI systems for content recommendation and moderation. The whistleblowers' claims indicate that these AI systems were deliberately configured or allowed to promote harmful content, causing harm to users and communities. This constitutes an AI Incident because the AI systems' use directly or indirectly led to harm through the spread of harmful content, fulfilling the criteria for harm to communities under the AI Incident definition.

The article explicitly discusses the use of AI-powered recommendation algorithms on social media platforms that have increased the spread of harmful content, leading to real and significant harms such as exposure to violence, hate speech, and harassment. The AI systems' development and use have directly contributed to these harms by prioritizing engagement over safety. This fits the definition of an AI Incident because the AI systems' use has directly led to harm to communities and user safety. The article does not merely warn of potential harm but documents ongoing harm and internal acknowledgment of these risks, confirming the incident classification.

The social media platforms use AI systems (content recommendation algorithms) that influence what content users see. The article details how these AI systems were used or allowed to promote harmful content, leading to actual harm to users (bullying, harassment, exposure to violent and hateful content). This constitutes harm to communities and potentially violations of user rights. The involvement of AI in content curation and the resulting harm meets the criteria for an AI Incident, as the AI systems' use directly led to harm. Therefore, this event is classified as an AI Incident.