OpenAI's GPT-6 Astra Shows Reduced Hallucinations, Vulnerable to Hidden Prompt Injections

2026-09-04

OpenAI's GPT-6 Astra model has demonstrated a decrease in hallucinations and effectively blocks most direct prompt injections. However, it remains susceptible to attacks embedded within documents it processes.

Source: The Decoder

Reported by VERA Newswire.

More from September 2026 in The Record.