The Decoder
techCenter
OpenAI's GPT-6 Astra hallucinates less but remains vulnerable to hidden prompt injectionstranslating…
OpenAI's GPT-6 Astra hallucinates less than its predecessor and blocks 99.99 percent of direct prompt injections. But when attacks are hidden inside documents the AI reads, the model still gets cracked in 8.5 percent of scenarios. Claude Opus 5 does better at 4.8 percent. For…
Keywords#GPT-6 Astra#OpenAI's GPT-6#percent#OpenAI's GPT-6 Astra#astra hallucinates
Comments
Sign in to join the discussion.
Loading comments…