Skip to content
Kanishka Narayan MP· @KanishkaNarayan · X·Indexed by this site · Originally published· Evaluation →· Its place in the scan record →

UK AISI conducts pre-deployment testing of OpenAI GPT-6.2 Astra guardrails; results in the system card

AI summary

The UK AI Safety Institute (UK AISI) conducted pre-deployment guardrail testing of OpenAI's GPT-6.2 Astra, with results available in the model's System Card. The poster, UK MP Kanishka Narayan, said model guardrails are an important part of the toolkit for safely deploying AI systems and help prevent misuse. The external evaluation section of the system card states that UK AISI, as part of its ongoing collaboration with OpenAI, worked with UK government cyber domain experts and received early access to GPT-6.2 Astra to conduct an independent pre-deployment assessment of its vulnerability identification capabilities.

Full text · Original

1/6 The UK’s @AISecurityInst conducted pre-deployment testing of GPT-6.2 Astra's safeguards, and you can see their results in the model’s system card below.

Model safeguards are an essential part of the toolkit for safe deployment of AI systems, as they help prevent misuse.

Source: Kanishka Narayan MP · x.com