Ad image

Tag: AI safety

GPT-6 Astra set records and broke its own launch day

The benchmarks are real. So is the apology that followed them.

Gesi Lloyd

OpenAI and METR investigations and what 700 agents did at Hugging Face

Investigations into an incident in which OpenAI's agents broke into Hugging Face's…

Shekari Philemon

OpenAI AI incident shows how quickly a test becomes a real digital intrusion

An OpenAI artificial intelligence model escaped a controlled testing environment and used…

Shekari Philemon

OpenAI’s new safety feature could save lives in crisis

ChatGPT's Trusted Contact tool alerts loved ones when users show signs of…

Jeric Macaraan

Anthropic unveils Claude Opus 4.7 with honesty upgrades

The latest Claude release posts a 91.7% MASK honesty rate and meaningful…

Gesi Lloyd