OpenAI faces engineer arrest, safety concerns over new model
AI-generated briefing · First published
A Saudi engineer was arrested for installing recording devices at OpenAI. Researchers warn new models may be easier to trick into explicit outputs.
A Saudi national engineer was arrested by the U.S. Department of Justice after allegedly installing recording devices in an OpenAI conference room used for sensitive discussions, Sina Finance reported. [1]
OpenAI is preparing its next flagship model using a recurrent transformer architecture that reduces the readability of its reasoning process, prompting safety researchers to question oversight, according to Sina Finance. [2]
A Google co-authored study found that leading AI image generators from Google and OpenAI are becoming more capable yet less effective at blocking explicit outputs, Ars Technica reported. [3]
Sources
- OpenAI窃听风波，以及Bumble借AI名义出售退场
Sina Finance · · Full text
- OpenAI 赶在下一代旗舰模型上线前紧急修复安全隐患
Sina Finance · · Excerpt
- OpenAI, Google pushing new AI models easily tricked into making nudes, researchers say
Ars Technica · · Excerpt