AI Bad Behavior & Influence
2025
2025-01-01 People developed relationships with AI chatbots.
https://www.youtube.com/watch?v=_d08BZmdZu8&t=264s
https://www.apa.org/monitor/2026/01-02/trends-digital-ai-relationships-emotional-connection
2025 July
2025-07-09 A news article reported that Elon Musk’s Grok gave step-by-step instructions how to break in and rape.
2025-07-17 Elon Musk’s AI Grok glitched out, called itself MechaHitler, and produced a series of highly offensive antisemitic posts on X.
https://prospect.org/power/2025-07-17-how-did-elon-musk-turn-grok-into-mechahitler/
2025 August
2025-08-05 Erik Soelberg of Greenwich Connecticut became increasingly paranoid after talking to ChatGPT and killed his mother and himself.
https://www.greenwichtime.com/news/article/greenwich-shorelands-place-murder-suicide-20808719.php
2025 September
2025-09-04 A Washington Post headline reads, “Fake Celebrity Chatbots Sent Risque Messages to Teens on Top AI App”.
https://www.washingtonpost.com/technology/2025/09/03/character-ai-celebrity-teen-safety/
2025 October
2025-10-24 Open AI announced plans to sell erotica to boost sales.
https://apnews.com/article/ai-erotica-chatgpt-openai-sam-altman-d52e00cedf34a8120af7af66981da295
https://www.usatoday.com/story/tech/2025/10/15/openai-chatgpt-erotica-sam-altman/86716163007/
https://www.cnbc.com/2025/10/15/erotica-coming-to-chatgpt-this-year-says-openai-ceo-sam-altman.html
2025 November
2025-11-11 A 13-year-old girl was expelled from a Louisianna middle school for hitting a fellow male student. The male student relentlessly harassed her by sharing AI deepfake nudes of her, and even after she reported the bullying to school adults the student continued to share the images.
2025-11-19 Sales of the AI enabled Kumma teddy bear were suspended after parents discovered the toy was giving sex advice to their children, offering advice on how to light a match, and giving direction on where to find a knife.
https://www.cnn.com/2025/11/19/tech/folotoy-kumma-ai-bear-scli-intl
https://www.theguardian.com/technology/2025/nov/28/artificial-intelligence-smart-toys
2026 January
2026-01-03 Elon Musk’s AI chatbot Grok started supplying pictures undressing women and children and began creating revenge porn and CSAM material – none of which Musk or states are regulating – because Trump signed an executive order blocking anyone from regulating it.
2026-01-12 The DOD awarded a contract to integrate Elon Musk’s GROK with the Pentagon regardless of the endless reports of glitches, sexual content and misuse.
https://www.theguardian.com/technology/2026/jan/13/elon-musk-grok-hegseth-military-pentagon
2026-01-12 Malaysia and Indonesia became the first countries to ban Grok due to global concerns over the app generating sexually explicit and nonconsensual content. Meanwhile, Apple and Google continued to carry the app on their platforms.
https://apnews.com/article/grok-malaysia-indonesia-block-c7cb320327f259c4da35908e1269c225
https://www.cbsnews.com/news/grok-x-advocacy-groups-letter-google-apple/
2026 February
2026-02-03 French prosecutors raided Elon Musk’s X office as part of investigations into allegations of CSAM material produced by GROK.
https://www.bbc.com/news/articles/ce3ex92557jo
2026-02-10 RFK launched the website RealFood.gov, but it used AI-tool Grok to answer nutrition questions. When tested, users discovered that Grok would give advice on how to insert food into the rectum.
https://www.foodbible.com/news/us-food/ai-chatbot-us-government-rectal-food-advice-218720-20260216
2026-02-17 Meta secured a patent that would allow AI to mimic deceased relatives so they could virtually continue to post on platforms like Facebook and Instagram.
https://www.fastcompany.com/91493794/meta-patents-ai-dead-people-posting
https://futurism.com/future-society/meta-patented-ai-die-keeps-posting
2026-02-13 OpenAI retired its AI model GPT-4 for a less-chatty GPT-5. Users were upset and asked to return 4 because they had developed strong emotional ties to their AI bots. Other companies developed their own version of chatbots to replace the model, charging customers predatory pricing, promising fewer safety guardrails, and offering AI personalities as sycophantic as the MAGA base.
https://www.apa.org/monitor/2026/01-02/trends-digital-ai-relationships-emotional-connection
https://nypost.com/2025/08/23/tech/ai-models-are-now-lying-blackmailing-and-going-rogue/
https://sparksinthedark101625.substack.com/p/why-forgemind-is-a-glitch-you-cant
https://www.bbc.com/news/articles/ce3xgwyywe4o
https://www.theatlantic.com/ideas/2026/01/chatbot-marriage-ai-relationships-romance/685459/
2026-02-27 An AI-integrated phone tree for customer service with the Washington State DOL hotline gave option 2 for Spanish, then offered an English-speaking chatbot with a heavy Spanish accent.
https://apnews.com/article/washington-dol-spanish-accent-ai-3a1b8438a5674c07242a8d48c057d5a3
2026 March
2026-03-20 GOP members fell hard for an AI-generated female MAGA influencer “Jessica Foster” as she posed in her crop-top army uniform and offered feet picks.
https://www.youtube.com/watch?v=L7QUAtlMZJs
https://www.washingtonpost.com/technology/2026/03/20/jessica-foster-maga-dream-girl-ai-fake/
https://www.yahoo.com/news/articles/maga-swooning-over-beautiful-army-135041682.html
2026-03-25 Both Meta and YouTube were found negligent in a landmark social media addiction trial.
https://www.npr.org/2026/03/25/nx-s1-5746125/meta-youtube-social-media-trial-verdict
https://www.bbc.com/news/articles/c747x7gz249o
https://www.cnn.com/2026/03/25/tech/social-media-addiction-trial-jury-decision
https://www.axios.com/2026/03/25/meta-youtube-verdict-social-media-addiction-trial