What happened
Copenhagen-based AI startup NobodyWho has secured €2 million in pre-seed funding to advance its mission to run Small Language Models (SLMs) locally on laptops and mobile devices, reducing reliance on cloud-hosted large models. The round was backed by PSV Tech, The Footprint Firm, and Norrsken Evolve.
NobodyWho’s approach focuses on privacy-preserving, cost-efficient, and energy-sustainable AI that executes directly on user devices rather than sending data to third-party cloud servers.
Who is affected
- Developers and product teams building AI features can integrate AI without deep machine-learning expertise or central cloud dependencies.
- Enterprises and organizations that handle sensitive data, particularly those in regulated industries, benefit from reduced risk of data exfiltration and an improved compliance posture.
- Security teams and CISOs accountable for data governance, risk, and infrastructure costs. Local inference shifts threat surfaces and cost models.
Why CISOs should care
- Data sovereignty and privacy: Running AI locally means sensitive information need not leave corporate devices, reducing exposure to cloud-side breaches or third-party misuse.
- Attack surface changes: Decentralized inference distributes computation away from central servers, potentially reducing risks associated with single points of failure in cloud endpoints.
- Cost and sustainability: Local SLMs can drastically reduce compute costs and carbon footprint compared with large cloud-hosted models, factors increasingly relevant to enterprise risk and sustainability portfolios.
3 practical actions for CISOs
- Assess local AI use cases: Identify where on-device AI could replace cloud AI in your organization to minimize data egress and strengthen compliance with policies such as GDPR.
- Update threat models: Incorporate device-level inference into security architecture reviews to track how distributed AI alters typical data flows and exposes potential vulnerabilities.
- Pilot with controls: Launch controlled pilots with frameworks like NobodyWho’s open-source SLM engine to measure performance, privacy benefits, and integration challenges before broad rollout.