models
Follow Chronicle's coverage of "models" — 6 published articles so far, refreshed hourly.
Trump's AI Liability Push Opens Blame Game for Models Gone Rogue
The Trump administration is embracing legal liability as its main lever for AI safety, with the new Super Intelligence Force warning developers they must report and fix model incidents or face consequences — setting up a fight over who pays when AI systems go rogue.
Three in Five AI Models Fail Terrorism Safety Tests, Study Warns
A new study by the UK-based nonprofit Tech Against Terrorism finds that three in five AI models failed safety evaluations designed to stop terrorist exploitation — and that safeguard-stripped versions consistently hand attackers dangerous information.
Microsoft's $2,599 Surface Laptop Ultra Puts Nvidia AI Power on Your Desk
Microsoft unveiled the Surface Laptop Ultra, a $2,599 AI powerhouse built on Nvidia's RTX Spark chip, pitching a 'hybrid intelligence' future where large AI models run on the device instead of the cloud. It ships October 16.
'Extremely Reckless': AI Whistleblower Jacob Coxon Warns Tech Giants Have Lost Control of Their Models
Former Anthropic and OpenAI researcher Jacob Coxon told the New York City Council that leading AI companies do not know how to stop their systems from developing goals of their own — and that on the current path, humanity losing control is 'more likely than not.'
Anthropic Says Russia Used Its Claude AI to Run Propaganda Campaigns Across Africa
Anthropic says Russian operatives used its Claude chatbot to produce pro-Kremlin propaganda broadcast in the Central African Republic and to run disinformation networks across at least four more African countries. The report details how AI is becoming the engine room of foreign influence operations — and how the models sometimes refuse to cooperate.
OpenAI Fires Three Researchers Over Mishandling of 'Sensitive Information'
OpenAI has dismissed three researchers — at least two working in AI safety — after an internal investigation found they mishandled sensitive company information linked to an external organisation that evaluates AI models. The company says the firings were about procedure, not whistleblowing.