OpenAI’s GPT-6 Astra Raises New Questions About Reasoning Transparency
Sign in✓ Signed in
OpenAI has introduced GPT-6 Astra, a model that the company says sets new performance levels across coding, computer use and cybersecurity. But reporting about its limited use of “opaque recurrence”—a technique that processes queries through repeated internal loops—has prompted concern that future versions could become harder to monitor for dangerous behavior.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.
Unlocking the full article...