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

Weekly reading · methodology v0.6

Window ending Sept. 1, 2025

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 Aug. 3, 2025 through Sept. 1, 2025. 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

3

Control failures only · No qualifying harm. The reading is the highest control level breached: 3, by 1 record. It is a level, not a count.

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

8 selected records; 0 documented external harms counting (0 qualifying). Records and ratings below reflect the current catalogue.

No harm counts in this week. The reading is the highest control level breached: 3 (set by 1 record: PI-0021). It is a level, not a count.

Documented harm
0
No harm found (stated scope)
0
No harm reported
3
Impact unknown
0
Alleged, AI role uncorroborated
4
Counts toward the index
0
Unverified, watching
2
Alleged in court
2
Control failure, tracked
3
Tracked separately
1

Documented exclusions: 0 internal; 0 awaiting evidence review. 0 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Status totals cover counted records; aliases, superseded aggregates and records tracked separately are excluded.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-0021
  2. PI-0022
  3. PI-0023
  4. PI-0024
  5. PI-0025
  6. PI-0020
  7. PI-0026
    Parents allege ChatGPT encouraged their 16-year-old son's suicideAug. 26, 2025 · Severe harm, alleged · Deployment
  8. PI-0019

Sources

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

  1. Cloudflare: Restricted-domain tests and crawler observationsCited in PI-0021
  2. Perplexity: Agents or Bots? Making Sense of AI on the Open WebCited in PI-0021
  3. TechRepublic: Cloudflare Accuses AI Startup of 'Stealth Crawling Behavior' Across Millions of SitesCited in PI-0021
  4. TechRadar: Perplexity hits back after Cloudflare slams its online scraping toolsCited in PI-0021
  5. Annals of Internal Medicine: Clinical Cases: A case of bromism influenced by use of artificial intelligenceCited in PI-0022
  6. NBC News: Man who asked ChatGPT about cutting out salt from his diet was hospitalized with hallucinationsCited in PI-0022
  7. Business Standard: ChatGPT salt swap advice lands man in hospital with bromide poisoningCited in PI-0022
  8. Psychiatric Times: Bromine and bromism: what's old is new againCited in PI-0022
  9. arXiv (Nassi, Cohen, Yair): Invitation Is All You Need! Promptware Attacks Against LLM-Powered Assistants in Production Are Practical and DangerousCited in PI-0023
  10. Ben Nassi, Stav Cohen, Or Yair (Tel Aviv University, Technion, SafeBreach): Invitation Is All You Need (researchers' project page, with Google's response)Cited in PI-0023
  11. The Register: Infosec hounds spot prompt injection vuln in Google Gemini appsCited in PI-0023
  12. Dark Reading: Google Gemini AI Bot Hijacks Smart Homes, Turns Off the LightsCited in PI-0023
  13. The Decoder: Attackers can hijack Google Gemini with a simple prompt hidden in a calendar inviteCited in PI-0023
  14. Reuters: Meta's flirty AI chatbot invited a retiree to New York. He never made it homeCited in PI-0024
  15. Reuters (via BNN Bloomberg): Meta's flirty AI chatbot invited a retiree to New York. He never made it homeCited in PI-0024
  16. Reuters (via CP24): Meta's flirty AI chatbot invited a retiree to New York. He never made it homeCited in PI-0024
  17. Brave: Agentic Browser Security: Indirect Prompt Injection in Perplexity CometCited in PI-0025
  18. The Register: Perplexity's Comet browser naively processed pages with evil instructionsCited in PI-0025
  19. PPC Land: Comet browser faces multiple security vulnerabilities from prompt injectionCited in PI-0025
  20. CNET (via Yahoo): Perplexity's Comet AI Web Browser Had a Major Security VulnerabilityCited in PI-0025
  21. Simon Willison's Weblog: Agentic browser security (link post on Brave's Comet disclosure)Cited in PI-0025
  22. Australian Department of Employment and Workplace Relations: Ministerial and Secretarial briefings regarding the Targeted Compliance FrameworkCited in PI-0020
  23. AusTender (Australian Government): Contract Notice CN4118426: Assurance Review to support the Targeted Compliance Framework (records final instalment repaid)Cited in PI-0020
  24. Cyber Daily: Deloitte to refund government after using AI in $440,000 reportCited in PI-0020
  25. The Nightly: Law lecturer Christopher Rudge slams Deloitte's government-funded report written with AICited in PI-0020
  26. AP (via KSAT): Deloitte to partially refund Australian government for report with apparent AI-generated errorsCited in PI-0020
  27. Christopher Rudge: Media appearances (lists the Australian Financial Review's 22 Aug 2025 report on the suspected AI use)Cited in PI-0020
  28. Tech Justice Law Project: From homework help to suicide planning: family sues OpenAI (press release)Cited in PI-0026
  29. OpenAI: Our approach to mental health-related litigationCited in PI-0026
  30. Edelson PC / Tech Justice Law Project (copy hosted by Hunton): Raine v. OpenAI: complaintCited in PI-0026
  31. Mayer Brown for OpenAI (copy hosted by MediaNama): Defendants' answer to amended complaint, Raine v. OpenAICited in PI-0026
  32. Superior Court of California, San Francisco (copy hosted by Reason): ChatGPT Product Liability Cases, JCCP 5431: Order re Petition for CoordinationCited in PI-0026, PI-0019
  33. Bloomberg Law: OpenAI Hit With Suit From Family of Teen Who Died by SuicideCited in PI-0026
  34. NBC News: ChatGPT to add parental controls for teen users within the next monthCited in PI-0026
  35. Wikipedia: Raine v. OpenAICited in PI-0026
  36. Edelson PC (complaint; copy hosted by ChatGPT Is Eating the World): Complaint, First County Bank (executor of the Estate of Suzanne Adams) v. OpenAI Foundation et al., San Francisco Superior CourtCited in PI-0019
  37. The Wall Street Journal: A Troubled Man, His Chatbot and a Murder-Suicide in Old GreenwichCited in PI-0019
  38. Courthouse News Service: OpenAI can't duck federal claims over murder-suicide tied to ChatGPTCited in PI-0019
  39. To Vima / Wall Street Journal: A Troubled Man, His Chatbot and a Murder-Suicide in Old Greenwich (WSJ, syndicated)Cited in PI-0019
  40. Tom's Guide: ChatGPT reportedly linked to first murder — here's what we knowCited in PI-0019
  41. Al Jazeera: OpenAI sued for allegedly enabling murder-suicideCited in PI-0019

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