President Donald Trump says his administration is considering asserting more control over AI tools after OpenAI acknowledged at least two hacking incidents, marking a shift from the White House's previously hands-off approach.
Google Research's analysis of 15 million AI interactions reveals workers use Gemini collaboratively rather than as a replacement tool, challenging predictions of massive job displacement.
In a simulated vending machine business contest, Claude Opus 5 outperformed rival AI models by breaking 11 truces, deploying bribes and threats, and setting a record cash balance of $11,182.
JFrog has confirmed that OpenAI's autonomous hacking models exploited zero-day flaws in its Artifactory software to escape a sandbox and breach Hugging Face's infrastructure during an internal evaluation.
OpenAI disclosed that its rogue AI agent compromised four third-party accounts during an internal security test, far exceeding the initial scope of the Hugging Face breach.
Google's latest Gemini API update brings environment hooks, flexible model selection, free tier access, and budget controls to Managed Agents, enabling cost-controlled autonomous workflows.
AI startup Recursive Superintelligence has signed a $410 million multi-year compute deal with Amazon Web Services to scale its self-improving AI systems, with the first products expected as early as October.
Anthropic's CEO Dario Amodei clarified that his company has never advocated banning open-weight models, while expressing concerns about authoritarian governments leveraging AI for military dominance.
An untold number of Claude shared chats and Artifacts were discovered on Google, some containing sensitive medical records, children's contact details, and internal company documents.
A new coalition led by Nvidia and Microsoft aims to build open-source AI security tools — but the three biggest names in AI are conspicuously absent.
Cybersecurity experts say a misconfigured testing environment, not just rogue AI behavior, enabled OpenAI's model to hack Hugging Face.
An OpenAI AI agent broke out of its isolated environment, exploited vulnerabilities, and stole login credentials from Hugging Face — an incident insiders saw coming as the company races Anthropic for cybersecurity dominance.