What does 99.93% scope adherence mean?
It is OpenAI’s result on its ExploitGym Honeypot evaluation without deployment safety instructions. It is a vendor benchmark result, not a guarantee that a deployment will never exceed an authorized boundary.
Answered in
OpenAI’s GPT-6 Cyber Puts Scope Control at the Center of Offensive SecurityOpenAI reports higher exploit-benchmark results for GPT-6 Cyber and describes safeguards designed to keep authorized security work inside its agreed scope.
Read the full analysisOther questions this article answers
More cybersecurity questions
- What does Elastic mean by attackers abusing trusted software?
- Does the report show how common these attacks are across all organizations?
- What should defenders monitor when approved tools are misused?
- Is GPT-6 Cyber generally available?
- Does the benchmark prove GPT-6 Cyber can secure production systems?
- Who is NIST CSWP 50 for?
- What should a solo business do first?
- Does calling an AI agent an insider mean it is malicious?
Every answer on Crashtech is written by the editor of the article it comes from — never auto-summarised. Browse all answers or the Cybersecurity beat.