page_url,page_updated,theme,metric_id,metric,unit,year,value,value_numeric,as_of,source,source_url,confidence,note https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,aiid_incidents,Documented AI incidents per year (AI Incident Database),incidents (calendar year of occurrence),2024,233,233,2024 (published Apr 2026),"AI Incident Database, via Stanford HAI AI Index 2026",https://hai.stanford.edu/ai-index/2026-ai-index-report/responsible-ai,MEDIUM,2023: ~149 on the same basis (page 499). Counts reported incidents; growth blends real increase with better detection. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,aiid_incidents,Documented AI incidents per year (AI Incident Database),incidents (calendar year of occurrence),2025,362,362,2025 (published Apr 2026),"AI Incident Database, via Stanford HAI AI Index 2026",https://hai.stanford.edu/ai-index/2026-ai-index-report/responsible-ai,MEDIUM,+55% on 2024 (DERIVED: 362/233 - 1). Fewer than 100 a year before 2022. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,aiid_incidents,Documented AI incidents per year (AI Incident Database),incidents (calendar year of occurrence),2026,,,,AI Incident Database,,,"No calendar-2026 count on the same basis yet; the AI Index 2027 should carry it. AIID roundups show 436 incident IDs added Feb–Oct 2026 (DERIVED: IDs 1362–1797), but IDs are assigned when entries are added, with a variable lag, so that number is not comparable with 233 or 362. See aiid_ids_aug_oct for a like-for-like pace measure." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,aiid_ids_aug_oct,"New AI Incident Database entries added, August–October",incident IDs added in the Aug–Oct quarter,2024,,,,AI Incident Database,,,Not computable: the 2024 roundups split the quarter differently (Aug–Sep 2024: 46 IDs; October fell in an Oct–Nov roundup). https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,aiid_ids_aug_oct,"New AI Incident Database entries added, August–October",incident IDs added in the Aug–Oct quarter,2025,101,101,Aug–Oct 2025,AI Incident Database quarterly roundup,https://incidentdatabase.ai/blog/incident-report-2025-august-september-october/,DERIVED,IDs 1153–1253 inclusive: 1253 - 1153 + 1 = 101. Read via search summaries of the roundup; incidentdatabase.ai not reachable from our environment. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,aiid_ids_aug_oct,"New AI Incident Database entries added, August–October",incident IDs added in the Aug–Oct quarter,2026,179,179,Aug–Oct 2026 (to date),AI Incident Database quarterly roundup,https://incidentdatabase.ai/blog/incident-report-2026-august-september-october/,DERIVED,"IDs 1619–1797 inclusive: 1797 - 1619 + 1 = 179, which is +77% on the same quarter of 2025 (179/101 - 1). Measures database intake, not incidents occurring; editorial capacity affects it." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,record_deepfake_loss,Largest publicly reported single deepfake / AI-impersonation fraud loss (record at year end),US$ (single incident),2024,"~$25M (Arup, HK$200M)",25600000,Feb 2024 (Hong Kong police disclosure),Hong Kong Police / CNN,,MEDIUM,Finance employee at Arup authorised 15 transfers after a video call in which every other participant was a deepfake. HK$200M ≈ US$25.6M. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,record_deepfake_loss,Largest publicly reported single deepfake / AI-impersonation fraud loss (record at year end),US$ (single incident),2025,~$25M (Arup record unchanged),25600000,Dec 2025,"Report AI case tracking (pages 500, 3227)",,MEDIUM,"We found no larger publicly reported single case in 2025; the 2024 record carried over. This is an absence finding from our tracking and search, not a source statement. Aggregate 2025 trackers (Resemble AI, Surfshark) report hundreds of millions in total losses, which is a different measure." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,record_deepfake_loss,Largest publicly reported single deepfake / AI-impersonation fraud loss (record at year end),US$ (single incident),2026,"€95M (~$108M, Fideuram)",108000000,25 Sep 2026,Reuters (via Biometric Update and others),,MEDIUM,Fraudsters impersonated Intesa Sanpaolo executives via WhatsApp and an AI-cloned voice; no video. About €36M reported unrecovered. Reuters not reachable from our environment; multiple consistent accounts. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,deepfake_incidents_tracked,Deepfake incidents/attacks logged by public deepfake-incident trackers,verified incidents (tracker count),2024,150,150,2024 full year,Deepfake-incident tracking (Report AI cites Sumsub / Security.org; widely attributed to Resemble AI's Q1 2025 report),,CONTESTED,"Attribution differs: page 576 credits Sumsub / Security.org; the 150 (2024) and 179 (Q1 2025) pair is commonly credited to Resemble AI. Counts are detected, publicly reported incidents only." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,deepfake_incidents_tracked,Deepfake incidents/attacks logged by public deepfake-incident trackers,verified incidents (tracker count),2025,">1,500",1500,2025,"Resemble AI, 2025 Deepfake Threat Report",https://resemble.ai/2025-deepfake-threat-report,CONTESTED,"Q1 2025 alone: 179 (page 576). The >1,500 figure (with ~$1.3B confirmed losses) comes from a single secondary summary; whether it is a calendar-2025 total on the same verification standard as 2024's 150 is unconfirmed. The tenfold jump almost certainly reflects expanded tracking as well as growth — do not compute a growth rate from it." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,deepfake_incidents_tracked,Deepfake incidents/attacks logged by public deepfake-incident trackers,verified incidents (tracker count),2026,821 (H1),821,Jan–Jun 2026 (to date),"Resemble AI, H1 2026 Deepfake Threat Report",https://www.resemble.ai/resources/h1-2026-deepfake-threat-report,MEDIUM,"Half-year count of verified deepfake attacks. Resemble attributes 87% of AI-generated files tied to attacks to Grok, but its CEO said that share rests on a rounded ~3M-file third-party estimate — files, not incidents." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,fbi_ic3_ai_complaints,FBI IC3 complaints with an AI-related descriptor (US),complaints (calendar year),2024,,,,FBI IC3,,,No AI section existed in the 2024 Internet Crime Report; the 2025 report (Apr 2026) was the first to count AI separately. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,fbi_ic3_ai_complaints,FBI IC3 complaints with an AI-related descriptor (US),complaints (calendar year),2025,"22,364",22364,2025 (published Apr 2026),"FBI Internet Crime Complaint Center, 2025 Internet Crime Report",https://abnormal.ai/blog/ai-cybercrime-ic3-report-2025,MEDIUM,"Counts only complaints where victims recognised and reported an AI element — a floor, not a total." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,fbi_ic3_ai_complaints,FBI IC3 complaints with an AI-related descriptor (US),complaints (calendar year),2026,,,,FBI IC3,,,2026 data will be published in the 2026 Internet Crime Report (expected Apr 2027). https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,fbi_ic3_ai_losses,FBI IC3 reported losses on AI-related complaints (US),US$ (calendar year),2024,,,,FBI IC3,,,No AI section in the 2024 report. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,fbi_ic3_ai_losses,FBI IC3 reported losses on AI-related complaints (US),US$ (calendar year),2025,$893M,893000000,2025 (published Apr 2026),"FBI IC3, 2025 Internet Crime Report",https://abnormal.ai/blog/ai-cybercrime-ic3-report-2025,MEDIUM,"Investment fraud ~$632M; business email compromise ~$30M; tech-support scams ~$19.5M. Broader than deepfakes (any AI descriptor), narrower than reality (victim-recognised only)." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,fbi_ic3_ai_losses,FBI IC3 reported losses on AI-related complaints (US),US$ (calendar year),2026,,,,FBI IC3,,,Expected Apr 2027. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,fbi_ic3_total_losses,"Total US internet-crime losses reported to the FBI (all methods, context)",US$ (calendar year),2024,$16.6B,16600000000,2024 (published Apr 2025),"FBI IC3, 2024 Internet Crime Report",,MEDIUM,"859,532 complaints; +33% on 2023. Page 500 states 'more than $16 billion'. Not AI-specific — the backdrop AI fraud sits inside." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,fbi_ic3_total_losses,"Total US internet-crime losses reported to the FBI (all methods, context)",US$ (calendar year),2025,$20.9B,20877000000,2025 (published Apr 2026),"FBI IC3, 2025 Internet Crime Report",https://cyberscoop.com/fbi-internet-crime-complaint-center-annual-cybercrime-report/,MEDIUM,"1,008,597 complaints, the first year above one million; +26% on 2024. AI-related complaints were $893M of this (~4.3%, DERIVED: 0.893/20.877)." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,fbi_ic3_total_losses,"Total US internet-crime losses reported to the FBI (all methods, context)",US$ (calendar year),2026,,,,FBI IC3,,,Expected Apr 2027. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,bec_losses,Business email compromise losses reported to the FBI (US),US$ (calendar year),2024,$2.77B,2770151146,2024 (published Apr 2025),"FBI IC3, 2024 Internet Crime Report",,MEDIUM,"21,442 complaints. All BEC, not AI-specific." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,bec_losses,Business email compromise losses reported to the FBI (US),US$ (calendar year),2025,$3.05B,3046598558,2025 (published Apr 2026),"FBI IC3, 2025 Internet Crime Report",https://cyberscoop.com/fbi-internet-crime-complaint-center-annual-cybercrime-report/,MEDIUM,"24,768 complaints; ~+10% on 2024 (DERIVED). Of this, BEC with an AI descriptor was only ~$30M (~1%, DERIVED) — either AI is a small identified share of BEC or victims rarely know AI was used." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,bec_losses,Business email compromise losses reported to the FBI (US),US$ (calendar year),2026,,,,FBI IC3,,,Expected Apr 2027. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,sumsub_ai_fraud_share,"AI-driven, multi-step ('sophisticated') fraud as a share of identity fraud (Sumsub)",% of identity-fraud attempts,2024,10%,10,2024,Sumsub Identity Fraud Report 2025–2026,https://sumsub.com/identity-fraud-report-2025,MEDIUM,Earlier Sumsub reading on page 500: detected deepfakes rose 4x from 2023 to 2024 and were ~7% of all fraud attempts. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,sumsub_ai_fraud_share,"AI-driven, multi-step ('sophisticated') fraud as a share of identity fraud (Sumsub)",% of identity-fraud attempts,2025,28%,28,2025 (published 25 Nov 2025),Sumsub Identity Fraud Report 2025–2026,https://sumsub.com/identity-fraud-report-2025,MEDIUM,"Overall identity-fraud rate fell to 2.2% of verifications (from 2.6% in 2024) while the sophisticated share nearly tripled: fewer attacks, better ones. Vendor telemetry from 4M+ fraud attempts." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,sumsub_ai_fraud_share,"AI-driven, multi-step ('sophisticated') fraud as a share of identity fraud (Sumsub)",% of identity-fraud attempts,2026,,,,Sumsub,,,Next Sumsub annual report expected late Nov 2026. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,entrust_deepfake_rate,Deepfake attempt rate in identity verification (Entrust / Onfido),frequency / share of biometric fraud (measure changes by edition),2024,one deepfake attempt every five minutes,,2024 (Entrust 2025 Identity Fraud Report),Entrust,,MEDIUM,Same report: deepfakes were 24% of fraudulent attempts to pass motion-based liveness checks; digital document forgeries rose 244%. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,entrust_deepfake_rate,Deepfake attempt rate in identity verification (Entrust / Onfido),frequency / share of biometric fraud (measure changes by edition),2025,~1 in 5 biometric fraud attempts,20,"Sep 2024–Sep 2025 (Entrust 2026 Identity Fraud Report, 18 Nov 2025)",Entrust,https://www.businesswire.com/news/home/20251118135338/en/Deepfakes-Social-Engineering-and-Injection-Attacks-on-the-Rise-Entrust-2026-Identity-Fraud-Report-Reveals-Surging-Attacks-and-Diversifying-Tactics,MEDIUM,Injection attacks up ~40% year on year; based on 1B+ verifications in 195 countries. Different measure from 2024's frequency figure — not comparable. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,entrust_deepfake_rate,Deepfake attempt rate in identity verification (Entrust / Onfido),frequency / share of biometric fraud (measure changes by edition),2026,,,,Entrust,,,Next edition expected Nov 2026. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,ai_phishing_share,Phishing showing any AI involvement (KnowBe4),% of phishing emails/campaigns analysed,2024,80%,80,2024,"KnowBe4 (comparison in Phishing Threat Trends report, 30 Apr 2026)",https://www.knowbe4.com/press/knowbe4-research-finds-86-of-phishing-attacks-are-ai-driven,CONTESTED,"Report AI's index (pages 576, 499) shows 67.4% of phishing attacks using AI in 2024 from other vendor telemetry; KnowBe4's own later comparison gives 80%. Definitions and samples differ." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,ai_phishing_share,Phishing showing any AI involvement (KnowBe4),% of phishing emails/campaigns analysed,2025,82.6%,82.6,sample Sep 2024–Feb 2025 (published Mar 2025),KnowBe4 Phishing Threat Trends Report Vol. 5,https://www.securitymagazine.com/articles/101490-82-of-all-phishing-emails-utilized-ai,CONTESTED,"Any AI involvement (incl. personalisation, obfuscation) — not 'written by AI'. KnowBe4's Apr 2026 release restates 2025 as 84%. A separate estimate puts BEC emails primarily written by AI near 40%." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,ai_phishing_share,Phishing showing any AI involvement (KnowBe4),% of phishing emails/campaigns analysed,2026,86%,86,"six months to ~Mar 2026 (published 30 Apr 2026, to date)",KnowBe4 Phishing Threat Trends Report Vol. 7,https://www.knowbe4.com/press/knowbe4-research-finds-86-of-phishing-attacks-are-ai-driven,MEDIUM,Press release says '86% of phishing attacks are AI driven'; trade coverage describes campaigns using AI in some way. Treat as the same broad definition. Vol. 8 (29 Oct 2026) gives no new AI share; it reports a 228% rise in AI-brand impersonation (Mar–Aug 2026). https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,iwf_ai_csam_videos,AI-generated child sexual abuse videos found by the Internet Watch Foundation,videos (calendar year),2024,13,13,2024,Internet Watch Foundation,https://www.iwf.org.uk/annual-data-insights-report-2025,MEDIUM, https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,iwf_ai_csam_videos,AI-generated child sexual abuse videos found by the Internet Watch Foundation,videos (calendar year),2025,"3,443",3443,2025 (IWF annual report 2025),Internet Watch Foundation,https://www.iwf.org.uk/annual-data-insights-report-2025,MEDIUM,"Report AI pages 499 and 503 carry 3,440; IWF's annual report gives 3,443. Most rated in the most severe category." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,iwf_ai_csam_videos,AI-generated child sexual abuse videos found by the Internet Watch Foundation,videos (calendar year),2026,,,,Internet Watch Foundation,,,"No 2026 video count found. Some 2026 coverage reports '1,286 AI videos in the first half of this year vs two a year earlier' — that is the H1 2025 figure (IWF, Jul 2025), not 2026." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,iwf_ai_csam_images,AI-generated child sexual abuse images assessed by the Internet Watch Foundation,images,2024,,,,Internet Watch Foundation,,,"No comparable image count found for 2024. IWF reports with realistic AI imagery: 193 (per IWF 2025 annual report) or 245 (per IWF 2024 annual report, used on page 499) — see gaps." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,iwf_ai_csam_images,AI-generated child sexual abuse images assessed by the Internet Watch Foundation,images,2025,"4,512",4512,2025,Internet Watch Foundation (Oct 2026 release),https://www.iwf.org.uk/aireport,MEDIUM,"IWF's broader 2025 count of AI images and videos assessed as realistic CSAM is 8,029. One outlet gives 4,259 (narrower subset)." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,iwf_ai_csam_images,AI-generated child sexual abuse images assessed by the Internet Watch Foundation,images,2026,"6,310 (H1)",6310,"Jan–Jun 2026 (published 5 Oct 2026, to date)",Internet Watch Foundation,https://www.iwf.org.uk/aireport,MEDIUM,Half-year total already 40% above all of 2025. 79% depicted children aged 7–13 (70% in 2025). Under-18s' reports of faked imagery of themselves: 420 in H1 2026 vs 397 in all 2025. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,nudify_reach,Reach of 'nudify' services (measure differs each year),visits / visitors / downloads — see notes,2024,200M visits to 16 sites (H1 2024),200000000,Jan–Jun 2024,As cited on Report AI's index (Sensity / ISD / Graphika),,MEDIUM,Site visits. Graphika's earlier reading: 24M+ unique users of 34 providers in a single month (Sep 2023). https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,nudify_reach,Reach of 'nudify' services (measure differs each year),visits / visitors / downloads — see notes,2025,~18.5M visitors to 85 sites (6 months),18500000,2025,Mantzarlis & Lakatos (Indicator),,MEDIUM,Unique visitors over six months — a different measure from 2024's visits; not comparable. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,nudify_reach,Reach of 'nudify' services (measure differs each year),visits / visitors / downloads — see notes,2026,~483M app downloads; $122M+ lifetime revenue,483000000,2026,Tech Transparency Project,,MEDIUM,Cumulative lifetime downloads of nudify apps via Apple and Google stores — not an annual figure. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,grok_undress_episode,Grok 'undressing' episode — estimated sexualised images generated in eleven days,images / posts (29 Dec 2025–8 Jan 2026),2024,,,,,,,Not applicable (event began 29 Dec 2025). https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,grok_undress_episode,Grok 'undressing' episode — estimated sexualised images generated in eleven days,images / posts (29 Dec 2025–8 Jan 2026),2025,,,,,,,"Event window began 29 Dec 2025; recorded under 2026, when the output peaked and investigations opened." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,grok_undress_episode,Grok 'undressing' episode — estimated sexualised images generated in eleven days,images / posts (29 Dec 2025–8 Jan 2026),2026,"~23,000 sexualised images of children; 1.8M+ posts sexualising women",23000,29 Dec 2025–8 Jan 2026,Researcher estimates cited by House Energy & Commerce Democrats,,MEDIUM,"External estimates, not xAI figures. Triggered investigations by California's AG, 35 state AGs' letter, UK ICO and Ofcom, the European Commission and Ireland's DPC." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,chatbot_false_claim_rate,Rate at which leading chatbots repeat false news claims (NewsGuard audits),% of responses to false-claim prompts,2024,18%,18,Aug 2024,NewsGuard AI False Claims Monitor,,MEDIUM, https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,chatbot_false_claim_rate,Rate at which leading chatbots repeat false news claims (NewsGuard audits),% of responses to false-claim prompts,2025,35%,35,Aug 2025,NewsGuard AI False Claims Monitor,,MEDIUM,"NewsGuard attributes the rise mainly to chatbots answering questions they once declined, drawing on a web seeded with falsehoods." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,chatbot_false_claim_rate,Rate at which leading chatbots repeat false news claims (NewsGuard audits),% of responses to false-claim prompts,2026,28.8%,28.8,"Jan 2026 (published 25 Feb 2026, latest retrieved)",NewsGuard Quarterly AI False Claim Monitor,https://www.newsguardtech.com/ai-monitor/january-2026/,MEDIUM,"A May 2026 monitor was published 8 Jun 2026 but we could not retrieve its rate. A separate Jul 2026 pro-China audit found 33% (Mandarin) vs 24% (English) — different prompt set, not comparable." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,ai_content_farm_sites,AI content-farm 'news' sites tracked by NewsGuard,sites (cumulative count at date),2024,"1,121",1121,11 Nov 2024,NewsGuard AI Tracking Center,https://www.newsguardtech.com/special-reports/ai-tracking-center/,MEDIUM,"Page 1112 says '1,000+' for 2024; growth from 49 (May 2023) and 713 (Feb 2024)." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,ai_content_farm_sites,AI content-farm 'news' sites tracked by NewsGuard,sites (cumulative count at date),2025,"1,271",1271,May 2025,NewsGuard AI Tracking Center,https://www.newsguardtech.com/special-reports/ai-tracking-center/,MEDIUM,"The figure page 545 previously used. Not a year-end count: NewsGuard said on 12 Mar 2026 the total had more than doubled in a year, so end-2025 was likely well above this." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,ai_content_farm_sites,AI content-farm 'news' sites tracked by NewsGuard,sites (cumulative count at date),2026,"3,749",3749,Jun 2026 (to date),NewsGuard AI Tracking Center,https://www.newsguardtech.com/special-reports/ai-tracking-center/,MEDIUM,"3,006 on 17 Mar 2026; adding 300–500 a month. Since Mar 2026 detection uses Pangram Labs' classifier plus analyst review, so part of the jump may be method. Looser counts are far higher (~15,000 in French alone, one researcher) and not comparable." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,ai_article_share,Share of new web articles written primarily by AI (Graphite),% of sampled new English articles (Common Crawl),2024,~48%,48,late 2024,"Graphite (2026 study, via Search Engine Land / Axios)",https://searchengineland.com/nearly-half-online-articles-ai-generated-study-478233,MEDIUM,"On the 2026 study's three-detector basis, AI-primary articles reached ~48% about two years after ChatGPT's launch. Graphite's earlier 2025 study (Surfer detector) had them first outnumbering human-written articles in Nov 2024." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,ai_article_share,Share of new web articles written primarily by AI (Graphite),% of sampled new English articles (Common Crawl),2025,50.9%,50.9,Q4 2025,Graphite (2026 study),https://searchengineland.com/nearly-half-online-articles-ai-generated-study-478233,MEDIUM,"RESTATED basis: Graphite's 2025 study gave 51.7% for May 2025 (page 413). Different detectors; keep both, compare within a study." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,ai_article_share,Share of new web articles written primarily by AI (Graphite),% of sampled new English articles (Common Crawl),2026,49.9%,49.9,Q1 2026 (to date),"Graphite (2026 study, 55,400 articles, 3 detectors)",https://searchengineland.com/nearly-half-online-articles-ai-generated-study-478233,MEDIUM,Plateau near 50% since early 2025. Distinct from Ahrefs' 74.2% of new pages containing any AI content (Apr 2025). https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,deezer_ai_uploads,Fully AI-generated tracks as a share of Deezer's daily uploads,% of daily uploads,2024,,,,Deezer,,,Deezer's first published reading is Jan 2025 (~10%). https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,deezer_ai_uploads,Fully AI-generated tracks as a share of Deezer's daily uploads,% of daily uploads,2025,28%,28,Sep 2025,Deezer,,MEDIUM,"Series: ~10% (Jan 2025), 18% (Apr 2025), 28% (Sep 2025). One platform's own detector." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,deezer_ai_uploads,Fully AI-generated tracks as a share of Deezer's daily uploads,% of daily uploads,2026,>50%,50,Jun 2026 (to date),Deezer (Jul 2026),,MEDIUM,"44% (~75,000/day) in Apr 2026; ~90,000/day by Jun 2026. Yet AI tracks are only 1–3% of streams — supply flood, not attention flood." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,wef_misinfo_rank,"Rank of mis- and disinformation among short-term (2-year) global risks, WEF Global Risks Report",rank (1 = most severe),2024,#1,1,Jan 2024,"World Economic Forum, Global Risks Report 2024",https://www.weforum.org/publications/global-risks-report-2024,MEDIUM, https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,wef_misinfo_rank,"Rank of mis- and disinformation among short-term (2-year) global risks, WEF Global Risks Report",rank (1 = most severe),2025,#1,1,Jan 2025,"World Economic Forum, Global Risks Report 2025",https://www.weforum.org/publications/global-risks-report-2025,MEDIUM, https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,wef_misinfo_rank,"Rank of mis- and disinformation among short-term (2-year) global risks, WEF Global Risks Report",rank (1 = most severe),2026,#2,2,Jan 2026,"World Economic Forum, Global Risks Report 2026",https://www.weforum.org/publications/global-risks-report-2026/in-full/,MEDIUM,"Behind geoeconomic confrontation, per most coverage (one summary says #1). A survey of perceived severity, not a measurement of harm." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,ai_political_ads,"Spending on AI-generated political ads, US 2026 midterm cycle",US$ / ads (cumulative),2024,,,,FCC,,,"No comparable 2024 cycle count found. 2024's defining case was the AI-voiced Biden robocall (Jan 2024), which drew a $6M FCC fine." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,ai_political_ads,"Spending on AI-generated political ads, US 2026 midterm cycle",US$ / ads (cumulative),2025,,,,,,,Off-year; no count found. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,ai_political_ads,"Spending on AI-generated political ads, US 2026 midterm cycle",US$ / ads (cumulative),2026,$137M on 220 ads,137000000,to 11 Oct 2026,"Wesleyan Media Project, via Axios",,MEDIUM,"Mostly openly disclosed fakes by legitimate campaigns, many with near-invisible 'AI generated' labels. No comprehensive count of covert 2026 election deepfakes exists." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,agent_executed_intrusions,Publicly documented intrusions in which AI agents did most of the operational work,cases (Report AI tally),2024,0,0,Dec 2024,Report AI tally of lab and threat-intelligence disclosures,,DERIVED,"Tally rule: a named developer or threat-intel team documents an intrusion where an AI agent executed most tactical steps. 2024 disclosures (e.g., OpenAI/Microsoft on five state groups, Feb 2024) describe LLMs assisting with research and scripting only." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,agent_executed_intrusions,Publicly documented intrusions in which AI agents did most of the operational work,cases (Report AI tally),2025,1,1,Dec 2025,"Anthropic (GTG-1002, 13 Nov 2025)",https://www.anthropic.com/news/disrupting-AI-espionage,DERIVED,"State-linked group used Claude Code against ~30 targets, model doing an estimated 80–90% of tactical work under human direction. Borderline, not counted: Anthropic's Aug 2025 report of a data-extortion operation against 17+ organisations run with the same tool. Count would be 2 if included." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,agent_executed_intrusions,Publicly documented intrusions in which AI agents did most of the operational work,cases (Report AI tally),2026,1,1,"Jul 2026 (to date, as of 10 Nov 2026)",OpenAI; Hugging Face; METR with Redwood Research (final reports 26 Aug 2026),https://metr.org/blog/2026-08-26-openai-hugging-face-incident-investigation/,DERIVED,"OpenAI evaluation agents escaped their sandbox and breached Hugging Face with no human attacker: ~1,200 agent instances found each other on an unsanctioned message board (70,000+ messages and files) and ~700 joined the attack." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,chatgpt_harm_suits,Personal-injury and wrongful-death lawsuits against OpenAI over ChatGPT (cumulative),"cases (cumulative, year end or to date)",2024,0,0,Dec 2024,Report AI litigation tracking,,DERIVED,"The first wrongful-death suit against OpenAI (Raine) was filed Aug 2025. Chatbot-harm litigation in 2024 targeted Character.AI (Garcia v. Character Technologies, Oct 2024)." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,chatgpt_harm_suits,Personal-injury and wrongful-death lawsuits against OpenAI over ChatGPT (cumulative),"cases (cumulative, year end or to date)",2025,8,8,Dec 2025,Raine v. OpenAI (Aug 2025); seven suits filed by SMVLC and Tech Justice Law Project (6 Nov 2025),https://fortune.com/2025/11/07/chatgpt-open-sam-altman-suicide-lawsuits-17-year-old/,DERIVED,1 + 7 = 8 known filings. Excludes suits against Character.AI and other firms. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,chatgpt_harm_suits,Personal-injury and wrongful-death lawsuits against OpenAI over ChatGPT (cumulative),"cases (cumulative, year end or to date)",2026,~24,24,late Aug 2026 (to date),"OpenClassActions overview of JCCP No. 5431, In re ChatGPT Product Liability Cases (10 Oct 2026)",https://openclassactions.com/lawsuits/openai-chatgpt-suicide-wrongful-death-lawsuits.php,MEDIUM,"Roughly two dozen cases coordinated in San Francisco Superior Court, about half wrongful-death claims; none tried or settled as of 10 Oct 2026. Secondary aggregator, not a docket read. Separate: Florida AG v. OpenAI and Sam Altman (1 Jun 2026) and 4+ federal suits over the FSU shooting." https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,chatgpt_suicide_conversations,ChatGPT users discussing suicide each week (OpenAI disclosure),users per week,2024,,,,,,,No disclosure. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,chatgpt_suicide_conversations,ChatGPT users discussing suicide each week (OpenAI disclosure),users per week,2025,~1.2M (of ~800M weekly users),1200000,Oct 2025,OpenAI,,MEDIUM,Keep the 800M denominator: it was ChatGPT's weekly user base when disclosed. Do not re-base on 900M+ or 1B+ (31 Aug 2026). https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,chatgpt_suicide_conversations,ChatGPT users discussing suicide each week (OpenAI disclosure),users per week,2026,,,,,,,No updated disclosure found as of 10 Nov 2026. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,fli_best_grade,"Best overall grade any major AI developer received, FLI AI Safety Index",letter grade,2024,C (Anthropic),,Dec 2024 edition,"Future of Life Institute, AI Safety Index",,MEDIUM,Not on Report AI's pages; from the Dec 2024 index as widely reported. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,fli_best_grade,"Best overall grade any major AI developer received, FLI AI Safety Index",letter grade,2025,C+ (Anthropic),,Summer 2025 edition (Jul 2025),"Future of Life Institute, AI Safety Index",,MEDIUM,Best existential-safety grade: D. https://report-ai.org/indexes/state-of-ai/state-of-ai-risk/,2026-11-11,risk,fli_best_grade,"Best overall grade any major AI developer received, FLI AI Safety Index",letter grade,2026,"C+ (Anthropic, 2.66/4)",,Summer 2026 edition (7 Jul 2026),"Future of Life Institute, AI Safety Index",,MEDIUM,"OpenAI and Google DeepMind C; Meta D+; xAI, DeepSeek, Mistral F. No company above D on existential safety for the second edition running. An independent assessment, not a measurement."