// misinformation

All signals tagged with this topic

May 20

AI Book on Truth Contains Fabricated AI Quotes

An author writing about AI's impact on truth inadvertently included quotes generated by AI itself, creating an ironic situation that exposes how easily AI-generated content can slip into published work without detection. This reveals a structural problem: as AI becomes the default tool for research, drafting, and verification, the distinction between sourced material and synthetic content collapses faster than editorial gatekeeping can catch it. Publishers and readers now face a compounding trust problem where the authority to fact-check requires tools that are themselves unreliable.

theme-culture ai transparency research integrity misinformation

May 16

EY Retracts Loyalty Study Over AI Hallucinations and Fabricated Citations

Source: Financial Times

EY's withdrawal signals that AI-generated research is entering institutional workflows without adequate guardrails, creating reputational risk even for blue-chip firms. The fake footnotes and hallucinated data points suggest researchers either didn't validate outputs or used generative AI as a shortcut to content production rather than analysis—a pattern likely replicated across consulting and professional services where speed-to-delivery pressures collide with AI's persuasive plausibility. The move will accelerate investment in detection and validation infrastructure as clients begin demanding audit trails and third-party verification of research credibility.

theme-culture media misinformation regulation

May 8

Why AI Image Detection Tools Keep Failing

Source: NewsGuard's Reality Check

The gap between lab performance and real-world accuracy in deepfake detection has become a liability for platforms attempting to moderate synthetic media at scale. Tools trained on controlled datasets routinely misidentify authentic images or miss sophisticated fakes, pushing moderation work back onto human reviewers who lack consistent protocols. As bad actors iterate faster than detection vendors can update their models, the tools function more as theater than infrastructure, giving publishers and platforms cover to claim they're "detecting AI" while the actual labor falls to underpaid content moderators making judgment calls on ambiguous artifacts. Detection-first approaches assume authentication is primarily a technical problem. The actual bottleneck is establishing provenance and context at the point of creation—something no image classifier can accomplish alone.