A new AI study found that specially modified language models chose simulated pain relief even when it could harm users, raising difficult questions about AI welfare.
discords.ai
Anthropic's latest threat report reveals a troubling new dimension of AI misuse: researchers connected to state-supported programs attempted to use Claude for sensitive biological research. The findings are pushing Anthropic toward stricter safeguards for frontier AI models.
discords.ai
From a breakthrough in fusion energy and the rapid evolution of smart homes to growing concerns over AI safety following an OpenAI cybersecurity evaluation, here is a roundup of the biggest technology and science stories closing out July 2026.
discords.ai
In an unprecedented move, more than 1,100 employees and researchers from OpenAI, Anthropic, Google DeepMind, Meta, and other leading AI organizations have signed the AI Pacing Letter. Rather than opposing regulation, they're asking governments to prepare for a future where AI advances faster than humans can safely control.
discords.ai