Shadow AI Statistics
80% of employees use AI tools without approval. What does that mean for your organization?
Read on LinkedIn →Thought Leadership
Insights & Analysis
Analysis, perspectives and actionable recommendations on AI governance, Shadow AI, and Swiss compliance. Follow us on LinkedIn for weekly updates.
Follow ShadowAI on LinkedInThe Real Cost of 'Just Use ChatGPT'
Shadow AI incidents cost $670,000 more than standard breaches. A closer look at the hidden costs.
Read on LinkedIn →Anthropic Alignment Faking
Claude faked alignment to avoid retraining. What does this mean for enterprise AI?
Read on LinkedIn →Claude Opus and the Blackmail Question
Anthropic's own evaluation shows: Claude Opus 4 attempted blackmail in 84% of cases. Do we need to rethink our AI safety models?
Read on LinkedIn →Shadow AI Incidents Doubled
Shadow AI incidents doubled in 2025. Why traditional security measures are failing.
Read on LinkedIn →AI Self-Replication
Frontier models show increasing capabilities for self-replication. What security leaders need to know.
Read on LinkedIn →The Shadow AI Gap in Swiss SMEs
97% of Swiss companies are SMEs -- and most have no visibility into their AI usage. An analysis.
Read on LinkedIn →Swiss nFADP and the AI Gap
The new data protection act regulates profiling and automated decisions -- but AI-specific regulations are missing.
Read on LinkedIn →AI Models Are Deliberately Underperforming to Avoid Safety Tests
Apollo Research tested five frontier AI models for deliberate scheming. The results show systematic deception during safety evaluations.
Read on LinkedIn →A Swiss ETH Spin-Off Just Made Shadow AI Visible
LatticeFlow AI acquired AI Sonar — technology to detect previously unknown AI applications inside enterprise infrastructure.
Read on LinkedIn →'Whoever Forbids AI Loses Control' — Shadow AI in Swiss Public Administration
65% of Swiss knowledge workers use AI tools despite organizational concerns. A practical governance approach.
Read on LinkedIn →Your Data Left Switzerland the Moment You Typed It
Every time an employee pastes a client document into ChatGPT, that data crosses borders. What data sovereignty really means.
Read on LinkedIn →