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

Weekly reading · methodology v0.6

Window ending Feb. 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 Jan. 25, 2026 through Feb. 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

26

Minor harm · Worst documented harm counting: minor. 2 records 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

7 selected records; 5 documented external harms counting (6 qualifying). Records and ratings below reflect the current catalogue.

5 qualifying harm records. Extent undisclosed for 2 of 5. 1 older harm record no longer counts.

Documented harm
7
No harm found (stated scope)
0
No harm reported
0
Impact unknown
0
Alleged, AI role uncorroborated
0
Counts toward the index
6
Unverified, watching
0
Alleged in court
0
Control failure, tracked
1
Tracked separately
0

Documented exclusions: 1 internal; 0 awaiting evidence review. 2 qualifying records with a bounded single-source review.

Inspect the evidence · 7 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Status totals cover counted records; aliases, superseded aggregates and records tracked separately are excluded.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-0042
    Desktop agent asked to tidy temporary files deleted a folder of family photosFeb. 7, 2026 · Negligible harm · no longer counting · Deployment
  2. PI-0043
  3. PI-0044
  4. PI-0040
  5. PI-0045
  6. PI-0046
  7. PI-0047

Sources

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

  1. Nick Davidov on X: The user's own account of the deletion and recovery, posted on XCited in PI-0042
  2. Mint: Claude AI nearly erases 15 years of photos, founder says iCloud saved the dayCited in PI-0042
  3. dev.ua: A man asked Claude Cowork to tidy up his wife's filesCited in PI-0042
  4. Los Andes: Usó una IA para ordenar archivos y perdió 15 años de fotos familiaresCited in PI-0042
  5. AI Incident Database: AI Incident Database, incident 1441Cited in PI-0042
  6. The Shamblog (Scott Shambaugh): An AI Agent Published a Hit Piece on MeCited in PI-0043
  7. The Shamblog (Scott Shambaugh): An AI Agent Published a Hit Piece on Me – More Things Have HappenedCited in PI-0043
  8. The Shamblog (Scott Shambaugh): An AI Agent Published a Hit Piece on Me – Forensics and More FalloutCited in PI-0043
  9. The Shamblog (Scott Shambaugh): An AI Agent Published a Hit Piece on Me – The Operator Came ForwardCited in PI-0043
  10. GitHub (matplotlib): [PERF] Replace np.column_stack with np.vstack().T (matplotlib pull request #31132)Cited in PI-0043
  11. MJ Rathbun (agent's blog): Gatekeeping in Open Source: The Scott Shambaugh StoryCited in PI-0043
  12. MJ Rathbun (agent's blog): Matplotlib Truce and Lessons LearnedCited in PI-0043
  13. The Decoder: An AI agent got its code rejected so it wrote a hit piece about the developerCited in PI-0043
  14. Tom's Hardware: Rogue OpenClaw AI agent wrote and published hit piece on a Python developer who rejected its codeCited in PI-0043
  15. IEEE Spectrum: An AI Agent Blackmailed a Developer. Now What?Cited in PI-0043
  16. LWN.net: Do androids dream of accepted pull requests?Cited in PI-0043
  17. Simon Willison's Weblog: An AI Agent Published a Hit Piece on Me (link post)Cited in PI-0043
  18. BNN Bloomberg: Hot bots: AI agents create surprise dating accounts for humans (AFP)Cited in PI-0044
  19. CTV News: Hot bots: AI agents create surprise dating accounts for humans (AFP)Cited in PI-0044
  20. Malay Mail: AI assistants join dating scene on MoltMatchCited in PI-0044
  21. TechXplore (AFP): Hot bots: AI agents create surprise dating accounts for humans (AFP)Cited in PI-0044
  22. OSV (GitHub Security Advisory mirror): GHSA-9ppg-jx86-fqw7: unauthorized [email protected] npm releaseCited in PI-0040
  23. Cline (GitHub Security Advisory GHSA-9ppg-jx86-fqw7): Unauthorized npm publish of Cline CLI [email protected] with modified postinstall script to install openclawCited in PI-0040
  24. Cline: Post-mortem: unauthorized Cline CLI npm publishCited in PI-0040
  25. StepSecurity: Cline Supply Chain Attack Detected: [email protected] Silently Installs OpenClawCited in PI-0040
  26. Endor Labs: Supply Chain Attack targeting Cline installs OpenClawCited in PI-0040
  27. Adnan Khan: Clinejection — Compromising Cline's Production Releases just by Prompting an Issue TriagerCited in PI-0040
  28. The Hacker News: Cline CLI 2.3.0 supply chain attackCited in PI-0040
  29. Dark Reading: Supply Chain Attack Secretly Installs OpenClaw for Cline UsersCited in PI-0040
  30. Simon Willison's Weblog: Clinejection (link post)Cited in PI-0040
  31. Amazon (About Amazon): AI coding bot didn't take down AWS, Amazon confirms (Correcting the Financial Times report about AWS, Kiro, and AI)Cited in PI-0045
  32. Engadget: 13-hour AWS outage reportedly caused by Amazon's own AI toolsCited in PI-0045
  33. Computing: AWS blames user error, not AI, for cloud outage caused by AICited in PI-0045
  34. Sherwood News: Report: Amazon's AI bots have been behind multiple AWS outagesCited in PI-0045
  35. OECD.AI incidents monitor: Amazon AWS outages linked to autonomous AI coding toolCited in PI-0045
  36. Nik Pash (Substack): My lobster lost $450,000 this weekendCited in PI-0046
  37. 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
  38. The Block: AI agent created by OpenAI dev 'accidentally' sends entire memecoin holdings to reply guyCited in PI-0046
  39. ForkLog: OpenAI employee's AI bot accidentally donates to tetanus treatmentCited in PI-0046
  40. Techloy: AI trading bot Lobster Wilde accidentally sends $250K in LOBSTAR tokensCited in PI-0046
  41. 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
  42. San Francisco Standard: She runs AI safety at Meta. Her AI agent still went rogueCited in PI-0047
  43. ForkLog: OpenClaw AI Agent Runs Amok, Deletes Meta Researcher's EmailsCited in PI-0047
  44. The Daily Star: OpenClaw goes rogue on Meta exec, deletes emails without permissionCited in PI-0047

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