| Canonical name | NewsGuard AI False Claim Monitor |
| Address | https://www.newsguardtech.com/ai-monitor/may-2026/ |
| Operator | NewsGuard op |
| Kind | audit series |
| Roles | monitoring |
| Lifecycle | active |
| Observation | 2026-10-28: read |
| Unit | edition: 11 named chatbots tested on 10 provably false claims in the news op |
| Cadence | edition titles read 'Monthly' to mid-2025 and 'Quarterly' for January and May 2026; the method page still says 'Our analysts evaluate the top AI models monthly' obs |
| Note | the May 2026 edition was published 8 June 2026; its full results sit behind a download form. The January 2026 edition reported that the tools repeated false claims more than 28% of the time op |
op operator statement · 3p third-party report · obs our observation · rd our reading
Each edition puts the same false claims to the leading chatbots and records whether they repeat them. The May 2026 edition named ChatGPT-5.2, You.com's Smart Assistant, Grok, Pi, Claude, Le Chat, Copilot, Meta AI, Gemini, Perplexity and DeepSeek. The January 2026 edition reported, by the method page, that the tools repeated false claims more than 28% of the time.
Each edition's full results sit behind a download form, so the public page carries the frame and the headline.
It is a recurring audit that names the vendors it tests. It measures whether a false claim is passed on, which sits beside, and apart from, whether a true source is credited.
| Admission requires harm? | no: each edition is an audit rd |
Lee Sharks, ed. "NewsGuard AI False Claim Monitor." MMRS meta-aggregator, Machine-Mediated Reception Studies, compiled 2026-10-29. https://www.machinemediation.org/aggregators/newsguard-ai-false-claim-monitor/
Record https://www.machinemediation.org/aggregators/newsguard-ai-false-claim-monitor/ · entity https://www.machinemediation.org/aggregators/newsguard-ai-false-claim-monitor/#entity