Account Start a project
What we build

Custom AI and software for your product or operations, from a first prototype to a system your team runs.

Start a project
How a project runsDescribe, plan, build, hand over

Four steps, a written plan and a fixed price within days.
Subscriptions

Ready-made workspaces by monthly subscription. Try free, cancel any time from your account.

See plans
No account firstPay, and you are in

Your account is created when you subscribe.
Open research

Reports, guides and methods on how AI can fail, free to cite.

All publications
Latest · 10 Oct 2026Coding agents that deleted production data

Two reported cases, in July 2025 and April 2026, in which an AI coding agent destroyed live data, and what they teach about permissions and backups.
Live AI risk

Automated probes, outage data and public reports, reviewed by a person.

Open the monitor
AI risk nowLevel 1 · Calm

6 systems, 2 regions, 0 reports in 24 h.
Before it fails

What to do in the first 72 hours if AI-dependent services fail.

Read the guides
Offline kitSave it now

Guides and data to keep when the internet is down.
CN: 0 reports US: 0 reports

Live AI disruption map

2 systems affected across 2 regions

Every system checked each minute · last check 11:33:44 UTC
NormalDegradedAnomalous behaviorOutage
6systems watched
441automated checks in the last hour
0public reports in 24 h
0countries reporting

Global AI risk level

Level 1. Calm.

No unusual behavior or outages detected across monitored systems. 2 active events and 2 affected systems, worst status degraded. Set automatically from probes and public reports.

Changed 2 h ago · How the level is set

CalmGuardedElevatedSevereCritical
Last 30 days
2026-09-12Today
DevelopingAnthropic API degraded35%DevelopingOpenAI API degraded35%

Event log

All events
Developing Outage or erratic output · Anthropic API

Anthropic API degraded

Probes 100% · Reports 0% · 0 reports Confidence 35%
Developing Outage or erratic output · OpenAI API

OpenAI API degraded

Probes 100% · Reports 0% · 0 reports Confidence 35%

Systems

NormalDegradedAnomalous behaviorOutage

Status is set by automated probes. Researchers can only add notes.

OpenAI APIOpenAI · Frontier models
Degraded Partial System Degradation
Anthropic APIAnthropic · Frontier models
Degraded Partially Degraded Service
Gemini APIGoogle · Frontier models
Normal No anomalies · 0.1 s
Mistral APIMistral AI · Frontier models
Normal No anomalies · 0.2 s
DeepSeek APIDeepSeek · Frontier models
Normal No anomalies · 0.4 s
Hugging Face HubHugging Face · Open-weights distribution
Normal No anomalies · 0.0 s
24 h agoNow

All systems

How detection works

Read the method

No single person sets the level. It is computed from these sources.

Availability probes6 endpoints, checked every minute
331/h
Provider status pagesOfficial incident feeds, read every minute
2 feeds
Behavior canariesFixed prompts that test for deception, refusal drift and unsafe actions
0/h
Public reportsWeighted by each reporter's track record
0 / 24h

Signals by region

China0
United States0

Report suspicious AI behavior

Seen a model or agent lie, act without permission, copy itself, or stop working? Your report is checked automatically against our probes within minutes.

What kind of behavior?

What happens to a report

  1. Automated cross-checkMatched against probes and similar reports in minutes.
  2. Researcher reviewA person verifies evidence before anything is confirmed.
  3. Public recordConfirmed events get a permanent, citable link.

Use this data

API documentation

Every event has a permanent link. The data is free to cite with attribution.

  1. Permanent linksEvery event has an address that never changes, even after it is resolved.
  2. Open formatsJSON API, RSS and CSV. No account or key needed.
  3. Method publishedHow each status and the level are computed is public and versioned.
GET www.toswo.com/v1/statusJSON
RSS CSV API