Loading the catalogue…
Loading the catalogue…
InternLM is developed by the Shanghai Artificial Intelligence Laboratory, a state-backed Chinese research institute, not a private commercial lab. Its models are broadly open-weight and actively maintained with a large open-source toolchain, but the organisation has not signed the EU AI Act's GPAI Code of Practice, publishes no discoverable GDPR-facing privacy policy, and — as a China-headquartered, government-linked entity — sits outside CLOUD Act/FISA reach but within China's own state data-access framework, a distinct sovereignty risk for EU regulated customers.
InternLM is produced by a Chinese state-backed research institute rather than a private commercial entity, meaning strategic direction and disclosure practices may be shaped by government priorities rather than commercial/regulatory incentives familiar to EU enterprises.
No evidence the organisation has signed the EU GPAI Code of Practice; major Chinese AI labs generally have not signed it, leaving compliance demonstration to alternative (unverified) means.
No discoverable privacy policy, DPA, or DPO contact on the creator's official domain, making GDPR posture unverifiable.
As a China-domiciled state-affiliated entity, data and models are subject to China's national intelligence and cybersecurity laws rather than CLOUD Act/FISA — a different but non-trivial sovereignty concern for EU regulated customers.
No published security certifications (SOC 2/ISO 27001), bug bounty, or disclosure policy could be found for the organisation.
Stav’s assessment
Editorial assessment, not legal advice. Stav's risk ratings, scores, and verdicts are our own analysis of publicly available information and may be incomplete or out of date. Verify independently before making compliance or procurement decisions.
Publishes a full open-source toolchain (training, deployment, evaluation, data-parsing tools) alongside model weights, supporting independent audit and reuse.
Active HuggingFace organisation with multiple actively maintained model families (InternLM, InternVL, Intern-S1) and ongoing dataset/paper contributions.
SHLAB is recognised by outside analysts as a leading technical contributor to AI safety research and benchmarking, including human-value-alignment and adversarial-robustness evaluation work.
Continues active model releases as of early 2026 (Intern-S1-Pro), indicating ongoing operation rather than contraction or wind-down.
Published safeguards & certifications