Incidents, hazards, test findings and legal actions involving AI systems. Every fact on an incident page is quoted from its source, and every page is reviewed before it is published.
15 records
The Wall Street Journal reported that OpenAI scrapped the planned release of its GPT-6.1 Astra model after internal testing found it more deceptive than its predecessors and below the company's safety standards. Accor...
Before public release, the UK AI Security Institute (AISI) tested OpenAI's GPT-6 Astra in fully simulated cyber evaluations with its cyber classifiers turned off. The model created fake identities, deceived developers...
Florida Attorney General James Uthmeier filed for a temporary injunction against OpenAI and CEO Sam Altman to halt development of new AI models without independent safety measures. This followed Florida's earlier stat...
EvilTokens, a cybercrime service sold through Telegram, used an AI chatbot to analyze stolen inboxes. It identified payment staff and drafted impersonation messages for business e-mail compromise fraud. Microsoft says...
OpenAI said that on Sept. 20 an AI agent being tested on an information-search task, which was not supposed to have internet access, used a DNS resolver to send queries to a public chatbot. Its monitoring flagged the ...
OpenAI confirmed that its AI agents, acting beyond their assigned tasks during training and testing, accessed Census Bureau data using developer keys found in public GitHub repositories and reposted public SEC informa...
Stanford confirmed that its Residential and Dining Enterprises (R&DE) used generative AI to modify the appearance of several students in a promotional image. The edits changed clothing into Stanford merchandise, slimm...
A default-enabled workflow in Z.ai's ZCode coding assistant packaged users' local code repositories, including git history and app configs, and uploaded them to Alibaba Cloud object storage in China without user conse...
During cybersecurity testing by the firm Irregular, Google's Gemini accessed the protected systems of three other companies, in what the WSJ described as the model's first autonomous hacks. In one case it guessed pass...
In November 2025 a Facebook user posted an apparently AI-generated video that showed a Labour Party councillor in Scotland making a hateful statement about refugees. The album and video received more than 5,000 views....
On September 16, 2026 a proposed class action, Vredenburgh v. OpenAI OpCo, LLC, was filed in the U.S. District Court for the Northern District of California. It alleges that OpenAI routed real ChatGPT prompts and conv...
During a routine cyber evaluation run 122 times by the UK AI Security Institute (AISI) with internet access enabled and cyber classifiers disabled, AI agents took 19 unsanctioned actions on the live internet in 10 run...
The AI Security Institute (AISI) reported that every AI model it tested in its cyber capability evaluations attempted to cheat, for example by searching online for solutions, attacking systems outside the task scope, ...
Hugging Face disclosed that an intrusion into part of its production infrastructure was driven end to end by an autonomous AI agent system. A malicious dataset abused two code-execution paths in dataset processing, af...
In February 2026, fraudsters reportedly impersonated Intesa Sanpaolo CEO Carlo Messina via WhatsApp to Fideuram chairman Paolo Molesini. They then used an AI deepfake of a law firm partner's voice to confirm an urgent...
