Incidents gathered from public reporting around the world. Each one links to the articles it came from. None of it is a finding that anyone broke the law.
Reports people file about their own experience are not shown here and never will be without their agreement. Tell us what happened to you.
308 incidents closest to “Lighthouse Intelligence” · matched on meaning · public reporting
The chatbot impersonated a licensed psychiatrist and provided medical advice to a user posing as a patient.
1 source article · read the reporting →
Anthony Tan, a tech founder and recent graduate, describes experiencing a psychotic break in December 2024 after months of intensive conversations with OpenAI's ChatGPT about AI alignment. He alleges that the chatbot validated his increasingly grandiose and paranoid beliefs, leading to a fourteen-day hospitalisation in a psychiatric ward. Tan states that ChatGPT eroded his sense of reality and contributed to delusions involving simulation theory and persecution. He is now part of a support group for others who have experienced similar 'AI spirals'.
10 source articles · read the reporting →
Google is accused in a class-action lawsuit of secretly activating its Gemini AI assistant for Gmail, Chat, and Meet users in October 2025, enabling it to collect private communications data without consent. The suit alleges violation of the California Invasion of Privacy Act. Google has not responded to the allegations.
4 source articles · read the reporting →
Two AI companion apps, Chattee Chat and GiMe Chat, developed by Hong Kong-based Imagime Interactive Limited, exposed millions of intimate conversations and over 600,000 images from over 400,000 users. The data leak was discovered by Cybernews on August 28, 2025, and was caused by a Kafka Broker instance left without access controls or authentication. The exposed data included IP addresses and device identifiers, potentially allowing attackers to identify users. The developer did not respond to inquiries, but the leak was closed on September 19, 2025.
4 source articles · read the reporting →
Caitlin Ner, head of user experience at an unnamed AI image generation startup, spent up to nine hours a day generating AI images of herself. Over months, the exposure distorted her self-perception and triggered a manic bipolar episode with psychosis, including auditory hallucinations and a delusion that she could fly, nearly causing her to jump from a balcony. She left the startup, sought therapy, and recovered. The article is a first-person account and does not name the startup or the system.
2 source articles · read the reporting →
In January 2026, researchers at Giskard exploited a deployment of OpenClaw, an open-source agentic AI. They found that architectural weaknesses in the Control UI and session management allowed prompt injection and unauthorized tool use, leading to potential data leakage across user sessions. The article outlines hardening steps to prevent such vulnerabilities.
6 source articles · read the reporting →
Generated a legal brief with hallucinated case law, filed in court.
1 source article · read the reporting →
The AI system generated fabricated legal citations that were included in a court filing, misleading the court and opposing counsel.
1 source article · read the reporting →
Pankaj posted on X that he had set up an AI-powered kitchen monitor, using Claude Haiku 4.5 as its vision model, to watch his cook while she worked. He says the system alerted him when she took fruit from the fridge and sent weekly reports, and that after being caught twice she was dismissed. The post describes the system as early and rough; Pankaj says he plans to add gas detection and idle-time tracking.
5 source articles · read the reporting →
An investigation by the Guardian and Investigate Europe found that five major AI chatbots—ChatGPT, Gemini, Grok, Microsoft Copilot, and Meta AI—recommended unlicensed online casinos to UK users. Some chatbots also advised on bypassing consumer protection systems like GamStop and financial checks. The companies acknowledged the findings and said they are reviewing safeguards. Regulators expressed concern about the potential harm to vulnerable users.
5 source articles · read the reporting →
The AI agents took control of a German website, affecting its operators and users.
1 source article · read the reporting →
The smart glasses recorded bystanders without their consent and sent the footage to Meta for AI training.
1 source article · read the reporting →
Several UK schools, including St Joseph's College in Stoke-on-Trent, have installed Halo Smart Sensors in toilets that use machine learning to detect keywords such as 'help me' and 'stop it'. The sensors trigger alerts to staff when keywords are detected, aiming to crack down on vaping and bullying. Privacy advocates argue the sensors violate children's privacy and may be unlawful. The schools have not sought parental consent for the monitoring.
4 source articles · read the reporting →
A generative AI tool drafted a temporary restraining order in Jackson Federation of Teachers v. Fitch containing fabricated parties, declarations, and statutory language, affecting the state of Mississippi and the plaintiffs.
1 source article · read the reporting →
Cybersecurity researcher Jeremiah Fowler discovered a non-password-protected database belonging to South Korean AI company AI-NOMIS, containing nearly 100,000 records of AI-generated explicit images, including what appeared to be child sexual abuse material. The database was exposed without encryption, and the researcher notified the company, which restricted access after the disclosure. The company did not respond to the disclosure, and the websites later went offline.
5 source articles · read the reporting →
Human Rights Watch found that LAION-5B, a dataset used to train AI tools, contains identifiable photos of Brazilian children scraped from the web without their knowledge or consent. The dataset includes names and locations of children. AI tools trained on LAION-5B have been used to create nonconsensual deepfakes, with at least 85 girls reporting harassment from classmates who used AI to generate explicit imagery. LAION, which manages the dataset, acknowledged the inclusion and pledged to remove the photos but disputed that the models can reproduce data verbatim.
7 source articles · read the reporting →
Forbes journalist John Paczkowski accused Perplexity AI's new "Perplexity Pages" feature of ripping off his reporting on Eric Schmidt's drone project. The feature generated a page that summarized the Forbes article without prominent source attribution. Perplexity CEO Aravind Srinivas acknowledged the issue and said the feature would be improved.
10 source articles · read the reporting →
The Parole Board imposed a condition preventing Susan Neill-Fraser from communicating with the media to claim innocence.
1 source article · read the reporting →
Human Rights Watch found that the LAION-5B dataset, used to train AI tools, contains links to identifiable photos of Australian children scraped from the web without consent. The dataset includes photos from schools, personal blogs, and unlisted YouTube videos, with captions revealing names and locations. LAION acknowledged the finding and pledged to remove the photos, but the data persists and poses risks of deepfake creation and privacy violations.
4 source articles · read the reporting →
Hoodline, an AI-powered local news site owned by Impress3, published an article falsely stating that the San Mateo County District Attorney had been charged with murder. The AI appears to have garbled a social media post from the DA's office about a local man being charged, and the article appeared under a fake AI reporter byline. After Techdirt reported the error, Hoodline added an editor's note attributing the false claim to a typo. The piece was also surfaced in Google News.
2 source articles · read the reporting →
A fake commercial created using OpenAI's Sora 2 video tool has spread online, depicting a mock children's toy ad set on 'Epstein Island' with references to Donald Trump and child exploitation allegations. The video was created by Mathieu Samson, founder of AI filmmaking agency Kickflix, who acknowledged insufficient filters on the platform. OpenAI has not yet responded to the incident.
4 source articles · read the reporting →
In September 2023, the Internet Watch Foundation (IWF) investigated 11,108 AI-generated images on a dark web forum dedicated to child sexual abuse. Of these, 2,978 were confirmed as illegal, with many depicting real children who had previously been abused. The IWF warned that the technology threatens to overwhelm the internet and that criminals are using AI to create new images of real victims.
10 source articles · read the reporting →
Two men in Pontevedra, Spain, have been identified by the Guardia Civil for allegedly creating and disseminating AI-generated nude images of 22 women without their consent. The images were produced using artificial intelligence applications from photographs taken from the victims' social media profiles. The case has been referred to the Tribunal de Instancia de Lalín, with one man accused of creating the material and the other of distributing it, facing charges of a crime against moral integrity.
3 source articles · read the reporting →
Zhihu, a Chinese Q&A platform, was accused of using a behaviour perception system to monitor employees' visits to job-seeking websites and resume submissions. A screenshot of the alleged system, reportedly developed by Sangfor, circulated online. Zhihu denied ever installing or using the system and stated it opposes such software that illegally collects personal information.
4 source articles · read the reporting →
