GPT-6 Astra Draws Scrutiny for Being Harder to Monitor Even as OpenAI Calls It More Aligned
4 Articles
4 Articles
(San Francisco = Yonhap News) Correspondent Kwon Young-jeon = With the emergence of OpenAI's new artificial intelligence (AI) model 'GPT-6 Astra,' which bills itself as Artificial General Intelligence (AGI)...
The company indicated that Astra would launch with “additional monitoring of the thought chain to detect and contain any action potentially without alignment”
ghacks.net
IT Security News - cybersecurity, infosecurity newsGPT-6 Astra Draws Scrutiny for Being Harder to Monitor Even as OpenAI Calls It More Aligned
The latest frontier model from OpenAI, GPT-6 Astra, is under review because of a compromise in its safety disclosures: while the company states that the model is more capable and better aligned, researchers seem to be losing some of their ability to see the model's written reasoning.OpenAI admits that Astra is more difficult to monitor via its chain of thought and has grown powerful enough to attain the company's top cybersecurity capability rat…
Coverage Details
Bias Distribution
- 50% of the sources lean Left, 50% of the sources lean Right
Factuality
To view factuality data please Upgrade to Premium
