License compliance with AI assisted coding
summary
Shows that AI assistants sometimes output verbatim copyrighted code — one example reaches a carbon copy after 11 regenerations including original comments — and that undetected snippets create legal, security and export risks. Presents Scan OSS (osskb.org) as an open-source scanner and fingerprint database that computes cryptographic hashes locally so code never leaves the machine, generates complete SBOMs (SPDX/CycloneDX, package URL/CPE), integrates with Fossology and commercial SCA tools, powers an Audit Workbench GUI and paid SLA APIs, and was used in a GPL case in Italy.
AI-written summary · about
chapters (8)
- 0:00AI and open source compliance
- 2:20AI disruption mirrors open source
- 4:46Copyright law and AI code
- 9:06Gray zones in AI code copying
- 13:18Undeclared components risks
- 16:26Open source tools for software bill of materials
- 21:02Scanning OSS for privacy
- 23:08Commercial solutions and summary
related talks
38mOpen Source Licensing in the Era of AI-Assisted Coding
14mDoes using AI code assistants involve OSS plagiarism risk?
16mAdvanced SAST: How to Secure AI-Generated Code
33mORT Automate compliance using Open Source & InnerSource
29mAI-Generated Code: Legal Risks and How to Reduce Them
33mAI-Powered Open Source Risk Management: ISO Self-Certification Kit and 5-Level AI Coding Guide
32mOpen Source Contributions in Today’s World
31mLicense Compliance 101