EDH · Excessive Data Handling

Explore 1 source-supported AI security cases for EDH · Excessive Data Handling in Google SAIF 2.0, with evidence, scope, and defensive analysis.

Risk category

An AI system collects, retains, processes, or shares user data beyond what the applicable policies allow. User data can include queries, conversation content, interactions, personalization settings, preferences, and models derived from that information. The resulting risk concerns inappropriate data use and retention, including policy or legal consequences, even when no outside attacker steals the data.

Official framework source ↗ · Open interactive explorer ↗

Research ·

Shadow AI and Enterprise Data Governance: What 22.4M Prompts Reveal

Harmonic's analysis of 22,458,240 enterprise GenAI prompts and file uploads shows that Shadow AI (employee use of unmanaged or unapproved AI tools) risk is concentrated but not simple: six apps produced 92.6% of detected sensitive exposure, and ChatGPT alone accounted for 71.2%. Cyberhaven, Verizon DBIR 2026, and IBM's 2025 breach-cost data point to the same control lesson: govern the data flow before source code, legal documents, financial forecasts, or regulated data leave through personal or unmanaged AI accounts.

Why it relates: Harmonic's analysis of 22,458,240 enterprise GenAI prompts and file uploads shows that Shadow AI (employee use of unmanaged or unapproved AI tools) risk is concentrated but not simple: six apps produced 92.6% of detected sensitive exposure, and ChatGPT alone accounted for 71.2%. The relevant category is Excessive Data Handling: An AI system collects, retains, processes, or shares user data beyond what the applicable policies allow.

Harmonic's analysis of 22,458,240 enterprise GenAI prompts and file uploads shows that Shadow AI (employee use of unmanaged or unapproved AI tools) risk is concentrated but not simple: six apps produced 92.6% of detected sensitive exposure, and ChatGPT alone accounted for 71.2%. Cyberhaven, Verizon DBIR 2026, and IBM's 2025 breach-cost data point to the same control lesson: govern the data flow before source code, legal documents, financial forecasts, or regulated data leave through personal or unmanaged AI accounts. The flow is ordinary work, not a hack. An employee uses a personal or unmanaged AI account to summarize, debug, translate, draft, or analyze work data. The prompt or upload leaves enterprise SSO, logging, retention, and data-use controls. Harmonic makes the exposure measurable. In 22,458,240 prompts and uploads across 665 tools, it found 579,113 sensitive exposures. Six apps caused 92.6%; ChatGPT alone caused 71.2%. Code, legal documents, and financial data made up 74.5% of exposed sensitive content. Personal accounts break governance. Cyberhaven found 32.3% of ChatGPT usage happened through personal accounts; Verizon found 67% used non-corporate AI accounts on corporate devices. Verizon also found source code was the most common data type submitted to external GenAI models in 858,440 DLP events. The cost is real. IBM reports one in five organizations had a breach due to shadow AI, and high shadow AI usage added an average $670,000 to breach costs. The defense is approved AI routes, data tags, pre-submit controls, and sanitized telemetry.

AIA · Read the full case ↗