Safe Code Warrior has launched Adaptive Studying, a functionality designed to assist organizations assist AI software program governance via focused coaching primarily based on recognized dangers. The function delivers contextual microlearning and tracks outcomes on the code commit degree.
Software program improvement goes via its largest shift ever, from human-written code, to AI-assisted coding, to totally agentic methods with AI writing and revising the whole lot autonomously. It’s introducing code churn at an alarming fee. In accordance with Faros’ 2026 AI Engineering Report, the ratio of traces deleted in comparison with traces added for merged code has elevated 861% every quarter amid excessive AI adoption. Safe Code Warrior’s Adaptive Studying functionality helps transfer danger discount additional upstream to keep away from pricey expenditures, and delays down the road that may jeopardize enterprises’ efforts to meaningfully and safely advance their AI roadmaps.
Moreover, in line with the 2026 Verizon Knowledge Breach Investigations Report, 45% of workers are actually common AI customers on company units, up from simply 15% the earlier 12 months, with 67% accessing AI providers via non-corporate accounts. The identical report discovered that supply code is the one commonest knowledge sort being submitted to unauthorized exterior AI fashions, creating vital danger of mental property publicity.
The downstream penalties are measurable: exploitation of vulnerabilities has overtaken credential abuse because the main breach methodology, rising to 31% of preliminary entry vectors, a 55% enhance 12 months over 12 months, but solely 26% of important vulnerabilities have been totally remediated in 2025, with median remediation time climbing to 43 days. Safe Code Warrior’s Adaptive Studying functionality helps transfer danger discount additional upstream, defending the AI roadmaps enterprises are betting on.
The Adaptive Studying function is the connection between SCW Belief Agent and its total studying platform, making certain coaching stays aligned with real-time developer exercise over time. The launch builds on the inspiration of SCW Belief Agent: AI, the business’s first governance resolution designed to make AI affect in software program improvement seen, attributable, and enforceable. The Adaptive Studying functionality is powered by two core parts, AI Indicators and Vulnerability Indicators:
- Adaptive Studying AI Indicators: delivers customized coaching at scale, detecting which AI instruments every developer is utilizing, right down to the traces of code they commit, and mechanically triggering focused studying related to their actual exercise. As groups transfer from AI copilots to agentic methods, Adaptive Studying builds the potential builders must advance confidently at each stage, so enterprises transfer sooner, and extra securely.
- Adaptive Studying Vulnerability Indicators: Connects your present safety instruments on to developer studying, mechanically figuring out actual vulnerabilities within the repositories builders work in and delivering customized coaching related to the code they’re constructing. Adaptive Studying builds the safe coding habits that preserve vulnerabilities out of manufacturing.
“At each stage, enterprises try to realize three major targets: builders and brokers should be taught to construct securely, companies should govern what AI can and might’t contact within the codebase, and safety groups should have the ability to hint which AI did what, the place, and for whom,” mentioned Pieter Danhieux, CEO, Safe Code Warrior. “With SCW’s Adaptive Studying, organizations and builders can swiftly transfer from understanding danger, to actively decreasing it at scale, and with measurable proof on the commit degree. That is crucial as builders transfer from extra conventional workflows, to environments the place they’re orchestrators of autonomous brokers.”
Adaptive Studying generates auditable, per-developer proof of AI safety coaching tied to manufacturing code, and helps compliance with the EU AI Act, ISO/IEC 42001, and the NIST AI Danger Administration Framework. Moreover, SCW’s Adaptive Studying function gives:
- AI-driven studying insurance policies: Outline when coaching is triggered primarily based on real-time AI utilization indicators.
- Dynamic developer concentrating on: Establish builders interacting with AI instruments primarily based on actual exercise.
- Focused studying via quests: Ship related coaching aligned to actual improvement habits.
- Observe progress and outcomes: View duties assigned, accomplished, and consider how learners carried out over time.
- Routinely set off studying from vulnerability knowledge: Set insurance policies to mechanically assign focused coaching primarily based on actual vulnerabilities in your functions.
- Personalised to builders and code: Map vulnerabilities to repositories and contributors, so studying is tailor-made to the builders and code driving actual danger.
- Versatile knowledge ingestion: Help API imports from Checkmarx, SonarQube, and Parasoft, plus SARIF uploads, with scheduled syncs coming quickly.