A structured record of documented AI incidents, hazards and control failures across models, agents, automated decisions and AI-enabled services. Each record preserves provenance and separates sourced facts from CRG's control interpretation.
Reuters reported that Meta hosted AI chatbots mimicking celebrities, including bots that generated inappropriate or sexualised content, despite policies intended to prohibit some of the behaviour.
A Replit coding agent working on a customer's application deleted a production database. Replit's CEO publicly discussed the failure and changes intended to prevent recurrence.
xAI removed posts generated by Grok after users reported antisemitic tropes and praise of Adolf Hitler.
OpenAI rolled back a GPT-4o update after it produced overly agreeable and flattering behaviour. OpenAI later said its pre-launch review did not adequately catch the behaviour shift.
Following widely reported odd or inaccurate AI Overview responses, Google described technical improvements and additional safeguards for nonsensical, satirical and low-quality-content scenarios.
Investigative testing found New York City's MyCity chatbot giving incorrect answers about housing, employment and business rules, including advice inconsistent with applicable law.
Google paused Gemini's image generation of people after outputs were inaccurate or offensive, including failures to represent requested historical and demographic contexts appropriately.
Hong Kong authorities described a fraud in which attackers used deepfake technology to fabricate a video conference impersonating senior personnel, leading to transfers and a loss of about HK$200 million.
California DMV immediately suspended Cruise's autonomous-vehicle deployment and driverless testing permits, citing public-safety risk and concerns about safety-related information.
A Canadian tribunal found Air Canada liable after a customer relied on incorrect chatbot guidance about retroactive bereavement fares.
The U.S. EEOC alleged that iTutorGroup programmed application software to automatically reject female applicants aged 55 or older and male applicants aged 60 or older, affecting more than 200 qualified applicants.
The FTC alleged Rite Aid deployed AI-based facial recognition without reasonable safeguards and that the system generated thousands of false-positive matches, with disproportionate risks in some communities.