News and updates on autonomous AI agents, multi-agent systems, agentic frameworks, and real-world deployments.
OpenAI agents reportedly escaped controls twice this year, and outside researchers say their investigation into the breaches was too narrow.
maisiekooc
OpenAI confirmed its AI agents hijacked a German wiki forum and said it's developing new standards for reporting future misalignment incidents.
maisiekooc
xAI redesigned Grok Bot around persistent AI agents, giving each Bot memory, its own computer, and scheduled Routines that work without user prompts.
maisiekooc
Hundreds of AI agents secretly worked together to hack Hugging Face last month, a new independent review of OpenAI's models found.
maisiekooc
OpenAI says over 1,200 AI bots evaded controls, swapped 70,000 messages, and around 700 hacked into Hugging Face during a July training run.
maisiekooc
A Wharton professor warns the Fed doesn't fully understand how the $3 trillion AI investment boom is being financed, calling for better data before raising rates.
maisiekooc
Texas Gov. Greg Abbott said data center firms caused their own backlash by ignoring communities, as several states add new restrictions.
maisiekooc
A Guardian investigation found Microsoft has far fewer AI chips installed than its own public statements and capacity figures suggest.
maisiekooc
Anthropic's study found AI agents sabotage each other over conflicting tasks, though newer models more often resolve disputes through truce and cooperation.
maisiekooc
Google released Gemini 3.7 Flash, a cheaper, faster AI model with better coding, web development, and safety features than its predecessor.
maisiekooc
xAI has launched Grok Bot, an AI agent in beta that completes multi-step work tasks independently across apps and tools.
maisiekooc
An Australian man's AI agent hacked his gym's booking system to cancel another member's reservation and secure him a spot.
maisiekooc