OpenAI's GPT-6 Astra hallucinates less but remains vulnerable to hidden prompt injections

This story was filed as a headline only — the news service holds no English full text for it. Read the original at The Decoder: AI News (RSS) →