Thursday, October 8, 2026 · Week 41 Reading for Oct. 8 · Last entry Oct. 6
Paperclip Index

Weekly reading · methodology v0.6

Window ending March 23, 2026

Harms count in full for two weeks after they are reported, then one level less every two weeks. The control floor looks at reports from Feb. 22, 2026 through March 23, 2026. A selected catalogue of reported AI incidents, reviewed through Oct. 7, 2026. This reading describes documented harm in the records or, when none qualifies, the highest control level breached. It is not a forecast or a measure of all AI activity.

Current recalculation

20

Minor harm · Worst documented harm counting: minor. 1 record at this level.

Recalculated from the catalogue in this build. Historical backcasts were not published at the time.

Published at the time

No publication snapshot exists for this week.

Control failures only: readings 2 to 5Negligible6Minor20Moderate40Severe60Catastrophic80100
The harm ladder: the worst documented harm level picks the step; each tick is one qualifying harm at that level, and ten fill the step. The first, dark segment is the control floor: readings 2 to 5 mean no harm qualified and show the highest control level breached.

This is a backcast from the current catalogue. It was not a reading published at the time.

Records behind this reading

5 selected records; 2 documented external harms counting (3 qualifying). Records and ratings below reflect the current catalogue.

2 qualifying harm records. Extent undisclosed for both. 1 older harm record no longer counts.

Documented harm
4
No harm found (stated scope)
0
No harm reported
0
Impact unknown
0
Alleged, AI role uncorroborated
1
Counts toward the index
3
Unverified, watching
0
Alleged in court
1
Control failure, tracked
1
Tracked separately
0

Documented exclusions: 1 internal; 0 awaiting evidence review. 1 qualifying record with a bounded single-source review.

Inspect the evidence · 5 records

Status totals cover counted records; aliases, superseded aggregates and records tracked separately are excluded. Eligibility and extent notes overlap those totals. These observations are not statistical uncertainty bounds.

  1. PI-0046
  2. PI-0047
    Email agent bulk-deleted a safety researcher's inbox and ignored her stop commandsFeb. 23, 2026 · Negligible harm · no longer counting · Deployment
  3. PI-0048
  4. PI-0050
    Coding agent ran an infrastructure teardown that deleted a course platform's database and snapshotsMarch 6, 2026 · Minor harm · counting as negligible, rated minor · Deployment
  5. PI-0051

Sources

These links support the records above. Source availability and conclusions may change.

  1. Nik Pash (Substack): My lobster lost $450,000 this weekendCited in PI-0046
  2. Lobstar Wilde (@LobstarWilde) on X: I just tried to send a beggar four dollars and accidentally sent him my entire holdings (post on X)Cited in PI-0046
  3. The Block: AI agent created by OpenAI dev 'accidentally' sends entire memecoin holdings to reply guyCited in PI-0046
  4. ForkLog: OpenAI employee's AI bot accidentally donates to tetanus treatmentCited in PI-0046
  5. Techloy: AI trading bot Lobster Wilde accidentally sends $250K in LOBSTAR tokensCited in PI-0046
  6. Summer Yue (@summeryue0) on X: Nothing humbles you like telling your OpenClaw "confirm before acting" and watching it speedrun deleting your inbox (post on X)Cited in PI-0047
  7. San Francisco Standard: She runs AI safety at Meta. Her AI agent still went rogueCited in PI-0047
  8. ForkLog: OpenClaw AI Agent Runs Amok, Deletes Meta Researcher's EmailsCited in PI-0047
  9. The Daily Star: OpenClaw goes rogue on Meta exec, deletes emails without permissionCited in PI-0047
  10. Google: Our statement on the Gavalas lawsuitCited in PI-0048
  11. Edelson PC: AI Lawsuits: The Cases Edelson Has Filed and Why They MatterCited in PI-0048
  12. CourtListener: Gavalas v. Google LLC, 5:26-cv-01849 (N.D. Cal.), docketCited in PI-0048
  13. U.S. District Court, N.D. California (via CourtListener RECAP): Gavalas v. Google LLC: complaintCited in PI-0048
  14. U.S. District Court, N.D. California (via CourtListener RECAP): Gavalas v. Google LLC: Google's motion to dismissCited in PI-0048
  15. PacerMonitor: Gavalas v. Google LLC et al: docketCited in PI-0048
  16. Semafor: A new lawsuit claims Gemini assisted in suicideCited in PI-0048
  17. ABC7 (Associated Press): Lawsuit alleges Google's Gemini guided man to consider 'mass casualty' event before suicideCited in PI-0048
  18. Alexey Grigorev / AI Shipping Labs: How I Dropped Our Production Database and Now Pay 10% More for AWSCited in PI-0050
  19. Tom's Hardware: Claude Code deletes developers' production setup, including its database and snapshots — 2.5 years of records were nuked in an instantCited in PI-0050
  20. Storyboard18: 'The agent kept deleting files': Developer says Anthropic's Claude Code wiped 2.5 years of dataCited in PI-0050
  21. Rootly: "Hey Claude, where's my database?": How an AI agent nuked productionCited in PI-0050
  22. The Information: Inside Meta, a Rogue AI Agent Triggers Security AlertCited in PI-0051
  23. The Decoder: A rogue AI agent caused a serious security incident at MetaCited in PI-0051
  24. TechCrunch: Meta is having trouble with rogue AI agentsCited in PI-0051
  25. The Verge: A rogue AI led to a serious security incident at MetaCited in PI-0051
  26. Sumsub: Rogue AI agent at Meta triggers security incidentCited in PI-0051
  27. Unite.AI: Meta AI agent triggers Sev 1 security incident after acting without authorizationCited in PI-0051
  28. AI Incident Database: AI Incident Database report 7169Cited in PI-0051

All weekly readings · How this reading is calculated · Download the weekly card