Google’s mid-tier Gemini 4 Neon ranks third on an independent index. Its safety results are mostly inherited, and its own report isn’t out yet.
Tag: AI Safety
Vals asked leading AI models to mark military targets in drone and satellite images. Most complied, and calling it a game broke many refusals.
GPT-6.2 Astra is cheaper and OpenAI’s first model rated High for self-improvement. Its own card says it now omits known problems more often.
OpenAI is embedding Apollo, METR, Transluce and Goodfire to test models during training. They can publish, but OpenAI can still redact.
Four nonprofits want a San Francisco court to stop five AI labs building anything stronger than today’s models. Their evidence: the labs’ own words.
IBM’s open Granite 4.1 LLMON models are trained to read data without obeying it. In IBM’s tests, injection attacks nearly stopped working.
Google’s cheap Gemini 4 Hydrogen roughly matches Argon on its own numbers. Its model card says it reached the cyber and CBRN alert thresholds.
Thinking Machines’ 594B open model Inkling 2 Small is on Hugging Face. In training it slipped a network block, and the lab says so openly.
Researchers gave each user their own AI agent. In all four tests the group did worse than with one shared agent. Simulated, but telling.
Zyphra’s open ZONOS3-Mini clones a voice from a clip of up to 60 seconds, in 43 languages, on a phone. Its docs say nothing about consent.