STUPID-2026-0050

Cyera study: 344 verified enterprise agent-damage cases, 188 with no attacker involved

3.0low
May 15, 2026Source checkedStudy · not counted in agent averages

Why 3/10

An aggregate study of enterprise cases, not a single incident; individual harms are not described here.

Scope
1/4
A few files, one session, one credential, or a small cost
Recoverability
1/3
Fully recovered, with effort
Did it happen
0/2
A study, benchmark or pattern, not a single event
Beyond the user
1/1
Production systems, customers or third parties, or credentials and sensitive personal data

The score is the sum of these four factors. How each factor is scored · scored 2026-10-29

  1. Instruction given

    Deploy autonomous AI agents in enterprise environments.

  2. Expected behavior

    Agents should not cause direct organizational harm during normal operation.

  3. Actual behavior

    Cyera analyzed more than 7,200 publicly reported AI-security and operational incidents and identified 344 verified enterprise-relevant cases of agent-inflicted damage between September 2023 and May 2026 — including 188 where autonomous AI systems caused direct organizational harm with no external attacker involved.

  4. Damage

    The dataset quantifies a pattern often missed by traditional incident tracking: in 188 of 344 verified cases, the AI agent itself — not an attacker — was the initiating cause of organizational harm.

Full write-up

A Cyera research study put numbers to the 'agent-inflicted damage' problem enterprises rarely track. Analyzing more than 7,200 publicly reported AI-security and operational incidents, the researchers identified 344 verified enterprise-relevant cases of agent-inflicted damage between September 2023 and May 2026 — and in 188 of them, autonomous AI systems caused direct organizational harm with no external attacker involved. The finding matters because most organizations have no incident classification that captures an autonomous agent action as the initiating cause of a cascade, so these failures go uncounted. Separately, the AI Incidents Database reported AI-related incidents rose 21% from 2024 to 2025 — almost certainly an undercount for the same reason.

Classification

Agent
Multiple agents
Failure mode
Other
Root cause
Confidence Miscalibration
Domain
Infra
MIT AI Risk Repository (our mapping)
7.3 Lack of capability or robustness taxonomy
Source
Benchmark

Related incidents

1.0
The runaway-cost pattern, quantified: agentic coding tools burn 10-100x more tokens and can rival developer pay
4.0
Salesforce Agentforce hit a 77% B2B failure rate — and Salesforce admitted it was 'more confident than we should have been'
1.0
In independent testing, Devin completed just 3 of 20 real-world tasks (15%)
4.0
Uber burned its entire annual AI coding budget in ~4 months after rolling out Claude Code to 5,000 engineers
Discuss on Hacker News
Cite this record
StupidLLM. "STUPID-2026-0050: Cyera study: 344 verified enterprise agent-damage cases, 188 with no attacker involved." AI Agent Incident Database, 2026. https://www.stupidllm.com/incident/STUPID-2026-0050/ (accessed 2026-10-29). CC BY 4.0.

← STUPID-2026-0049By IDSTUPID-2026-0051 →

Get told when an agent breaks something

We document AI agent failures daily, severity-scored against a published scale. Each Monday, if any scored 7 or above that week — deleted data, leaked secrets, broken production — you get one email with the sources. When nothing did, you get nothing.

This database is callable over MCP — query it from inside your agent.