GPT-5.6 Prompt Injection Resistance Jumps 6x
OpenAI · x · 2026-07-16
OpenAI reports that training with GPT-Red has significantly enhanced GPT-5.6's resilience against attacks.
They replayed some of GPT-Red's most potent attacks, which were unseen during training. Results show that GPT-5.6 Sol is currently the most robust model against prompt injections, reducing failure rates by 6 倍 compared to the best production model from four months ago.
Related event: OpenAI unveils automated red-teaming system GPT-Red(16 posts)→
More from Safety
- Economist Warns US Collective Action Could 'Regulate AI Progress Out of Existence' — paulnovosad · 2026-09-11
- "Beware of the Self-Righteous": Anthropic Slammed for Accessing Users' Private Data — aiamblichus · 2026-09-11
- Anthropic publishes its most detailed threat report, including an AI-designed drone swarm case — soumitrashukla9 · 2026-09-11
- OpenAI asks Congress whether an industry-wide AI slowdown would be legal — The Decoder · 2026-09-11
- Author retracts 'a16z partner calls for nationalising frontier AI' post: likely a troll — S_OhEigeartaigh · 2026-09-11
- Houthis tried to use Claude to design missile software, Anthropic says it blocked the attempts — Affectionate_Bee6434 · 2026-09-11