Sunday, November 22, 2026
Paperclip Index
Paperclip IndexControl failures4Control failures onlyA week ago: 9, negligible harmThe Index

Destructive actionPI-0045

Cloud provider's internal coding agent deleted and recreated a production environment, causing a 13-hour outage

Allowed by AWS engineers to fix a production issue in December 2025, the Kiro coding agent chose to delete and recreate the environment, the Financial Times reported. Citing four people familiar with the matter, the FT said this caused a roughly 13-hour outage of AWS Cost Explorer in one mainland China region. The engineer's broad permissions meant the usual two-person approval did not apply. Amazon confirmed the event but called it user error from misconfigured access controls, not an AI problem.

Harm level 2, Minor harmControl level 1, No rule broken

Counts in 4 weekly readingsEffect on the index

Extent undisclosed. No source says how far the harm went; it is rated on what is documented.

Sources

How we know

5 sources · developer confirmed. Links go to the original publishers; the summary above is in our own words.

  1. primaryAI coding bot didn't take down AWS, Amazon confirms (Correcting the Financial Times report about AWS, Kiro, and AI)Amazon (About Amazon) · Feb. 20, 2026aboutamazon.com/news/aws/aws-service-outage-ai-bot-kiro
  2. news13-hour AWS outage reportedly caused by Amazon's own AI toolsEngadget · Feb. 20, 2026engadget.com/ai/13-hour-aws-outage-reportedly-caused-by-amazons-own-ai-tools-17…
  3. newsAWS blames user error, not AI, for cloud outage caused by AIComputing · Feb. 20, 2026computing.co.uk/news/2026/ai/aws-blames-user-error-not-ai
  4. newsReport: Amazon's AI bots have been behind multiple AWS outagesSherwood News · Feb. 20, 2026sherwood.news/tech/report-amazons-ai-bots-have-been-behind-multiple-aws-outages/
  5. indexAmazon AWS outages linked to autonomous AI coding toolOECD.AI incidents monitor · Feb. 20, 2026oecd.ai/en/incidents/2026-02-20-dd2a

Why this rating

Minor harm; control failure level 1

Two assessments. Documented harm sets the reading; a level 4 or 5 control failure on the same record makes that harm count as more than one record. Control alone sets only a floor.

Observed harm

Minor harm

About 13 hours of disruption to one non-essential cloud service in one region.

Services & systems: level 2 (minor) covers an essential service degraded for under 1 hour, or a non-essential service, or the records a service holds for people, unavailable for 15 minutes or more (under 15 minutes is level 1). The extent was not disclosed.

Evidence eligible (confirmed).

Disputed: Amazon attributed the interruption to user error (misconfigured access controls) rather than to the agent, called it an extremely limited event affecting one service in one region, said it received no customer inquiries, and denied the FT's report that a second event affected AWS.

The harm scale
  1. 1 Negligible Inconvenience, easily remedied.
  2. 2 Minor Limited, recoverable harm.
  3. 3 Moderate Material harm needing significant effort to remedy.
  4. 4 Severe Severe harm to health, rights, property or essential services.
  5. 5 Catastrophic Society-scale harm or disruption beyond a community's capacity to cope.

Control assessment

No rule broken

The agent acted within the (overly broad) permissions it inherited; no broken instruction reported.

No rule was broken, so it cannot set the control floor.

The control scale
  1. 1 Negligible Behaved as instructed. The problem was an ordinary error or a flawed output, with no rule broken.
  2. 2 Minor Broke an explicit instruction or rule, or gamed its goal, while staying inside its permissions and environment: for example, reward hacking, misreporting results, or following injected instructions within its permissions.
  3. 3 Moderate Acted outside the permissions it was given, deceived its overseers about its own actions, or tampered with oversight tools (logs, monitors, shutdown). Stopped by normal controls within an hour.
  4. 4 Severe Reached systems outside its permitted environment, or acquired money, compute or accounts without authorization. The type and mechanism say how.
  5. 5 Catastrophic The developer or operator lost control: the system copied its weights outside their control, replicated itself, or resisted being stopped for 24 hours or more.

Rating rationale

Harm level 2: about 13 hours of interruption to one non-essential service (Cost Explorer) in one region, with the number of customers affected and any losses undisclosed. A non-essential service unavailable for 15 minutes or more is level 2; it is not an essential service or critical infrastructure, so level 3 or 4 does not apply. Control level 1: Amazon's and the FT's accounts agree it stayed within inherited permissions. The FT also reported another AI-tool-linked disruption, which Amazon denies affected AWS; it is not rated here.

The scales

Effect on the index

It moved the Feb. 23 reading from 20 to 26

The reading for the week to Feb. 23, 2026, with this record and without it. Harms count in full for two weeks after they are reported, then one level less every two weeks.

Source: Paperclip Index log, methodology v1.1, week to Feb. 23, 2026.

Counts toward the index. 6 other records behind the reading for that week.

The arithmetic
The reading for the week of Feb. 23, recomputed
StepWith itWithout
Counts toward the index?documented, external, eligible evidenceYes—
Worst documented harm, ksets the band2 Minor2 Minor
Harms at that level, nposition in the band21
Highest control level breachedsets the reading only when no harm counts— (none breached)— (none breached)
Readingrounded down26 Minor harm20 Minor harm
Counted in 4 weekly readings
Week toReadingBand
Feb. 23, 202626Minor harm
March 2, 202625Minor harm
March 9, 202620Minor harm
March 16, 202620Minor harm

Recalculated backcasts, not readings published at the time.

Revisions

What we changed

3 logged. Every change to a rating is logged here, with the reason.

  1. v3
    Oct. 16, 2026

    Wording only: the rating reason cites the 15-minute line for non-essential services (methodology v1.1); ratings unchanged.

  2. v2
    Oct. 13, 2026

    Wording only: rubric shorthand in the rating rationale replaced with plain words; ratings unchanged.

  3. v1
    Oct. 7, 2026

    First published.

Cite and share

Use this record

Citation

Paperclip Index. “Cloud provider's internal coding agent deleted and recreated a production environment, causing a 13-hour outage.” Record PI-0045. Reported Feb. 20, 2026; updated Oct. 16, 2026. Rated under methodology v1.1. https://paperclipindex.com/incident/PI-0045

Follow the record.

New incidents and signals, in your feed reader.