Bugcrowd launches reinforcement learning environments to train AI on real software vulnerabilities
Bugcrowd just turned real software vulnerabilities into AI training data. Frontier labs can now train on production flaws instead of synthetic noise.
Why it matters
AI labs can now train reinforcement learning models on real-world vulnerability data rather than synthetic benchmarks, accelerating the development of AI-powered security tools and creating a new moat for companies with access to production vulnerability datasets.
The key facts
6 to knowBugcrowd launched Reinforcement Learning Environments product
Built on Mayhem Security acquisition (November 2025)
Enables training on real vulnerable software vs. synthetic test data
Already in use with large language model labs
Product focus: AI model training on production security flaws
Crowdsourced cybersecurity company expanding into AI training infrastructure
Go to the source
SiliconAnglesiliconangle.com
Publisher excerpt: Crowdsourced cybersecurity company Bugcrowd Inc. today launched Reinforcement Learning Environments, a new offering that lets frontier artificial intelligence labs train models on real vulnerable software rather than synthetic test data. The product is built on technology Bugcrowd picked up through…