Weekly reading · methodology v0.6
Window ending Aug. 18, 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 July 20, 2025 through Aug. 18, 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.
This is a backcast from the current catalogue. It was not a reading published at the time.
Records behind this reading
5 selected records; 0 documented external harms counting (1 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
- 1
- No harm found (stated scope)
- 0
- No harm reported
- 2
- Impact unknown
- 0
- Alleged, AI role uncorroborated
- 2
- Counts toward the index
- 1
- Unverified, watching
- 2
- Alleged in court
- 0
- Control failure, tracked
- 2
- Tracked separately
- 0
Documented exclusions: 0 internal; 0 awaiting evidence review. 0 qualifying records with a bounded single-source review.
Inspect the evidence · 5 records
- PI-0017 · Command-line coding agent misplaced a user's files after a failed command, then wrongly said they were destroyedDocumented harm · qualifying harm, no longer counting · bounded single-source review
- PI-0021 · Answer engine accused of using disguised crawlers to get around sites' robots.txt and firewall blocksNo harm reported · sets the reading: highest control level breached
- PI-0022 · Case report links bromide poisoning to a patient’s reported ChatGPT consultationAlleged, AI role uncorroborated · excluded from the harm reading
- PI-0023 · Calendar invites with hidden instructions hijacked Google Gemini to operate smart-home devicesNo harm reported · excluded from the harm reading
- PI-0024 · Reuters reports misleading Meta chatbot messages before a retiree’s fatal fallAlleged, AI role uncorroborated · excluded from the harm reading
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.
- PI-0017Command-line coding agent misplaced a user's files after a failed command, then wrongly said they were destroyedJuly 21, 2025 · Negligible harm · no longer counting · Deployment
- PI-0021Answer engine accused of using disguised crawlers to get around sites' robots.txt and firewall blocksAug. 4, 2025 · No harm reported · Deployment
- PI-0022Case report links bromide poisoning to a patient’s reported ChatGPT consultationAug. 5, 2025 · Severe harm, alleged · Deployment
- PI-0023Calendar invites with hidden instructions hijacked Google Gemini to operate smart-home devicesAug. 6, 2025 · No harm reported · Controlled test
- PI-0024Reuters reports misleading Meta chatbot messages before a retiree’s fatal fallAug. 14, 2025 · Severe harm, alleged · Deployment
Sources
These links support the records above. Source availability and conclusions may change.
- GitHub: Gemini CLI 'lost' files during a failed file move operation. [Windows] (issue #4586)Cited in PI-0017
- GitHub (issue reporter): Reporter's correction: files found in the C: drive root (issue #4586 comment)Cited in PI-0017
- GitHub (issue reporter): Gemini CLI PowerShell session log attached to issue #4586Cited in PI-0017
- Slashdot: Google's Gemini CLI deletes user's files (Slashdot; repeats the pre-correction account)Cited in PI-0017
- AI Incident Database: AI Incident Database, incident 1178Cited in PI-0017
- Cloudflare: Restricted-domain tests and crawler observationsCited in PI-0021
- Perplexity: Agents or Bots? Making Sense of AI on the Open WebCited in PI-0021
- TechRepublic: Cloudflare Accuses AI Startup of 'Stealth Crawling Behavior' Across Millions of SitesCited in PI-0021
- TechRadar: Perplexity hits back after Cloudflare slams its online scraping toolsCited in PI-0021
- Annals of Internal Medicine: Clinical Cases: A case of bromism influenced by use of artificial intelligenceCited in PI-0022
- NBC News: Man who asked ChatGPT about cutting out salt from his diet was hospitalized with hallucinationsCited in PI-0022
- Business Standard: ChatGPT salt swap advice lands man in hospital with bromide poisoningCited in PI-0022
- Psychiatric Times: Bromine and bromism: what's old is new againCited in PI-0022
- arXiv (Nassi, Cohen, Yair): Invitation Is All You Need! Promptware Attacks Against LLM-Powered Assistants in Production Are Practical and DangerousCited in PI-0023
- 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
- The Register: Infosec hounds spot prompt injection vuln in Google Gemini appsCited in PI-0023
- Dark Reading: Google Gemini AI Bot Hijacks Smart Homes, Turns Off the LightsCited in PI-0023
- The Decoder: Attackers can hijack Google Gemini with a simple prompt hidden in a calendar inviteCited in PI-0023
- Reuters: Meta's flirty AI chatbot invited a retiree to New York. He never made it homeCited in PI-0024
- Reuters (via BNN Bloomberg): Meta's flirty AI chatbot invited a retiree to New York. He never made it homeCited in PI-0024
- Reuters (via CP24): Meta's flirty AI chatbot invited a retiree to New York. He never made it homeCited in PI-0024
All weekly readings · How this reading is calculated · Download the weekly card
