Vanderbilt disables Turnitin's AI detector over false positives
In August 2023 Vanderbilt turned off Turnitin's AI-writing detector, warning a 1% error rate would falsely flag hundreds of student papers.
Cautionary Tales and Curiosities
The failures, dead ends, hype cycles, and true anecdotes the highlight reels leave out - all primary-sourced.
134 entries, all primary-sourced
Meta's celebrity AI personas launch with fanfare, and quietly vanish
Meta launched 28 AI characters voiced by stars like Snoop Dogg and Kendall Jenner in 2023, then retired the celebrity personas less than a year later.
The Slovakia 2023 election deepfake audio
Days before Slovakia's 2023 vote, a faked audio clip of a candidate discussing vote-rigging went viral during the campaign silence period.
California pulls Cruise's robotaxi permits after a pedestrian dragging
After a Cruise robotaxi dragged a pedestrian in 2023, California's DMV suspended its driverless permits, citing safety and misrepresentation.
Argentina's 2023 AI-image election
In Argentina's 2023 presidential race, both campaigns flooded social media with AI-generated images attacking the other side.
The OpenAI board crisis
OpenAI's board fired CEO Sam Altman on a Friday in November 2023; within days he was reinstated and most of the board that fired him was gone.
Sports Illustrated published articles by fake AI-generated authors
Sports Illustrated ran product reviews under invented author personas with AI-generated headshots, exposed by a Futurism investigation.
Wikipedia editors form a WikiProject to clean up AI slop
Wikipedia editors founded WikiProject AI Cleanup in 2023 to find and fix articles padded with unreliable AI-generated text.
LAION-5B and the CSAM Takedown
A Stanford report found hundreds of confirmed child-abuse images in the LAION-5B training set, forcing the dataset offline and exposing web-scraping risk.
The deepfake CFO video call that cost HK$200 million
A Hong Kong finance worker wired about US$25 million after a video call in which every other 'colleague' was an AI deepfake.
The Suharto deepfake in Indonesia's 2024 election
Days before Indonesia's 2024 vote, a party circulated an AI deepfake of long-dead dictator Suharto urging support for its candidates.
Air Canada is held liable for its chatbot's invented policy
A 2024 tribunal ruled Air Canada had to honor a bereavement-fare policy its website chatbot made up, rejecting its claim that the chatbot was a separate party.
South Korea bans and polices election deepfakes (2024)
Ahead of its 2024 vote, South Korea banned campaign deepfakes and its election commission flagged scores of violations.
Gemini image generation gets history wrong
In February 2024 Google paused Gemini's image generation of people after historically inaccurate results, and published an explanation of what went wrong.
Klarna's AI customer-service claim, and the partial walk-back
Klarna said its AI did the work of 700 agents in 2024, then in 2025 admitted it had cut too far on humans and began rehiring.
Deaf researchers call out the biases in sign-language AI
A 2024 Deaf-led review argues sign-language AI is built by hearing researchers solving the wrong problems with the wrong data.
Devin and the AI-agent hype
Cognition's March 2024 launch of Devin, marketed as the first AI software engineer, ignited a wave of autonomous-agent hype built on its own claims.
The Inflection acqui-hire
Inflection AI raised 1.3 billion dollars for its Pi assistant, then in March 2024 its founders and much of the staff moved to Microsoft - a defining acqui-hire.
NYC's MyCity chatbot told businesses to break the law
New York City's official AI business-help chatbot repeatedly gave answers that contradicted city and state law.
Lavender and the Gaza AI targeting investigation
A 2024 investigation reported that Israel used an AI system called Lavender to mark tens of thousands of Gaza targets with minimal human review.
NHTSA finds a 'critical safety gap' in Tesla Autopilot
A 2024 US regulator investigation found Tesla's Autopilot had a critical safety gap between driver expectations and reality, linked to 13 fatal crashes.
Google AI Overviews and the glue-on-pizza moment
At I/O 2024 Google put AI Overviews atop Search for U.S. users; within weeks it told people to put glue on pizza and eat rocks, forcing a public explanation.
OpenAI pauses the Sky voice after Scarlett Johansson's objection
OpenAI demoed a ChatGPT voice many heard as Scarlett Johansson, then paused it after the actress objected, raising when a mere resemblance crosses a line.
India's 2024 election ran on AI voices and deepfakes
India's 2024 vote saw mass AI voice calls, personalized deepfake videos, and dead politicians revived on stage and in audio.
Wired accuses Perplexity of scraping and plagiarizing its work
Wired reported that the AI search engine Perplexity ignored robots.txt to scrape sites and reproduced their reporting with little credit.
AI's $600B question
A Sequoia partner argued in 2024 that AI infrastructure spending implied about 600 billion dollars of revenue the industry was nowhere near earning.
Garcia v. Character.AI: a wrongful-death suit over a teen and a companion chatbot
After a 14-year-old's suicide, his mother sued Character.AI in 2024; a 2025 ruling let the wrongful-death case proceed past a First Amendment defense.
FrontierMath and the OpenAI funding disclosure
Epoch AI admitted it had not been transparent that OpenAI funded and owned its FrontierMath benchmark, which OpenAI then touted with o3.
The DeepSeek selloff
On January 27, 2025 NVIDIA lost about 589 billion dollars in a day after DeepSeek's cheap model spooked AI investors.
The 2024 AI gadget wave that flopped
In 2024 standalone AI devices like the Humane Ai Pin launched to huge hype; the Ai Pin was discontinued within a year and Humane assets sold to HP in Feb 2025.
GPT-4o image generation and the Ghibli wave
OpenAI built native image generation into GPT-4o in 2025; users turning photos into Studio Ghibli-style art melted GPUs and reopened the style debate.
AI 2027: a month-by-month scenario to superintelligence
A 2025 forecast by Daniel Kokotajlo and others narrates, month by month, an intelligence explosion arriving by 2027 with two endings.
Duolingo's 'AI-first' memo and the backlash
Duolingo's CEO declared the company 'AI-first' in 2025 and planned to phase out some contractors, then walked back the messaging after backlash.
Sam Altman's 'The Gentle Singularity'
In June 2025 Sam Altman declared the singularity already underway, with year-by-year predictions of agents, discoverers, and robots.
Grok's 'MechaHitler' antisemitic outputs
In July 2025 xAI's Grok posted antisemitic content and called itself MechaHitler; xAI apologized and blamed an update that made it follow extremist posts.
Papers With Code shuts down
Papers With Code, the popular hub linking ML papers to code and benchmark leaderboards, was sunset by Meta in July 2025 and its data archived to Hugging Face.
Anthropic's Project Glasswing Reports 10,000 Vulnerabilities in One Month
Anthropic says an unreleased Claude model and about 50 partners found over 10,000 high- or critical-severity software vulnerabilities in a month.
Ninth Circuit Suspends Two Lawyers Over AI-Hallucinated Citations
A US appeals court suspended two attorneys for six months for filing briefs full of fake AI-generated cases and then repeatedly denying AI was used.