helpnetsecurity.com·19 november 2026
AI agents inherit employee access, and nobody's watching how they use it - Help Net Security

AI Adoption in Practice 2026
AI agents
AI gateways
AI tools
Anthropic
data loss prevention
data sharing
employee credentials
Google Workspace
Nudge Security
OAuth grants
OpenAI
Salesloft Drift breach
Security Risks
security teams
helpnetsecurity.com
AI agents built by employees often run on the creator's credentials, carrying their access to company data. The AI Adoption in Practice 2026 report from Nudge Security found that 44% of organizations have more than 10 AI agents, running on 15 different platforms. These agents, often missed by endpoint tools, can complicate audit trails and pose security risks. The report advises security teams to inventory agents and review OAuth grants to mitigate these risks. OpenAI and Anthropic integrations were found in 91% of organizations, with data often shared via prompts for routine tasks.

