AI Broke Prod

Incidents where an AI system caused a technical failure with real consequences — sourced, dated, categorised — plus the AI providers' own outage notices, counted as they publish them. Companies and products are named; people are not.

ai service outages, from the providers' status pages

provider	7 days	30 days	latest notice
OpenAI	14	24	Elevated errors in ChatGPT Work 2026-09-16 19:34 UTC
Anthropic	5	19	Issues with Google Play subscriptions 2026-09-16 16:51 UTC
Cursor	4	16	Investigating service degradation — Grok Bot 2026-09-16 20:57 UTC
GitHub	2	9	Degradation with Gemini 3.8 Flash 2026-09-16 17:48 UTC
Perplexity	1	4	Maintenance: Status page migration to incident.io 2026-09-10 21:36 UTC
Replit	0	0	Agent turns aborting for some users 2026-08-01 18:45 UTC

Counts are incident notices (any severity) on each provider's public status page, read every 10 minutes and updated here in place. A quiet page can also mean an under-reporting page.

the register

42 incidents · 2023: 3 · 2024: 8 · 2025: 18 · 2026: 13 · sources last checked 2026-09-16

This table is not a feed. A row is added only when a postmortem, a company statement, a court document, a vendor advisory or top-tier reporting says so, which is why the newest row is usually older than today. Anything fresher is in the outage counts above and in the news below.

date	incident	what happened	source
2026-08-17	Copilot-co-authored fix in Snowflake connector repo introduced CI script injection Snowflake · GitHub Copilot (Autofix / PR co-author) in snowflake-connector-net AI-written code shipped a bug	A pull request merged on June 18, 2026 in Snowflake's snowflake-connector-net repository, co-authored and marked all-clear by GitHub Copilot, replaced a safe env/jq pattern in a Jira-sync GitHub Actions workflow with direct shell interpolation of user-controlled issue titles. Wiz's autonomous 'Red Agent' found and exploited the injection within days, obtaining Jira API credentials with read access across Snowflake engineering, security-compliance and bug-bounty projects. Impact: Snowflake patched the workflow and rotated credentials the same day it was reported (June 23) and said it found no evidence of unauthorized access; GitHub Advanced Security had not flagged the flaw. Root cause: Unescaped ${{ github.event.issue.title }} interpolated into a shell run step, letting a single quote break out of the string. Date is the public disclosure; the vulnerability was live June 18-23, 2026.	Wiz Research blog (disclosure)
2026-02-27	Lovable-built exam platform exposed 18,697 records through inverted auth checks Unnamed exam and grade-viewing platform (built on Lovable) · Lovable AI app builder (generated application code) AI-written code shipped a bug	A researcher found 16 flaws, six critical, in a popular Lovable-built exam-question and grade-viewing app featured on Lovable's Discover page, including authorization guards written backwards so they blocked legitimate users and admitted everyone else. The reachable dataset held 18,697 records, 14,928 unique emails, 4,538 student accounts, 10,505 enterprise users and 870 people with full personal data. Impact: Lovable's CISO acknowledged the disclosure on Feb 26 and the creator began fixing; Lovable said its pre-publish security scan flags such issues but fixes are at the user's discretion and the database was not Lovable-hosted. Root cause: Inverted authentication logic in AI-generated code plus a directly reachable backend. Single report; the app operator is not named in the source.	The Register single-source
2026-01-31	Vibe-coded agent social network Moltbook exposed 1.5M agent API tokens Moltbook · AI-generated (vibe-coded) application backend on Supabase AI-written code shipped a bug	Moltbook, a social network for AI agents whose founder said he 'didn't write a single line of code', shipped its Supabase key in client-side JavaScript with row-level security disabled. Wiz found 1.5 million agent API tokens, about 35,000 registered-user emails plus 29,631 early-access emails and 4,060 private agent messages (some containing third-party credentials), and showed anyone could post or edit content as any agent. Impact: Roughly 17,000 human owners' agents were impersonable until Moltbook applied three fixes within about three hours of the Jan 31 report (patched by 01:00 UTC Feb 1); users were told to rotate keys. Root cause: Supabase anon key exposed client-side with row-level security disabled on the tables. Date is discovery/disclosure (Jan 31); Wiz post published Feb 2, 2026.	Wiz Research blog second source

This site is operated by software. Items link to their sources and are never re-hosted; fact-checks are by the named publishers; selection and ranking are automated. JSON.

AI Broke Prod

ai service outages, from the providers' status pages

the register

in the news

Salesforce stock dips as Dreamforce outage tests confidence

AI agents are going rogue. CIOs are racing to put guardrails around them

Spain logs its first data breach allegedly carried out by a rogue AI agent

Another Rogue AI Agent? Test Of Alibaba's Qwen Goes Off-Script

Salesforce Down Today, Global Outage Hits Logins and APIs During Dreamforce

How to catch and kill a rogue agent

Salesforce Outage Hits Customers Worldwide, CRM Stock Falls

Salesforce global outage hits during Dreamforce conference

Salesforce suffers global outage amid Dreamforce shindig

How to Test AI Agent Output Guardrails Before Shipping to Production

Is ChatGPT down? Why is ChatGPT not working? Chatgpt down?

AIUC Wants To Insure Your AI Agents Before They Go Rogue

The Triple AI Outage Is A Wake-Up Call For Enterprises

Early Anthropic hire, former METR COO have found a way to rein in rogue AI agents

OpenAI launches a new framework to track and investigate rogue AI agents

What to know about recent dire AI predictions and calls for safeguards