Adsafety.net vs NewsGuard

AI Made Fake Content Cheap — Designers Must Fix Trust

This analysis argues that generative AI has collapsed the cost of producing plausible but false content, intensifying Brandolini’s 'bullshit asymmetry'—where refuting false claims takes far more effort than making them. Production can now be parallelized and nearly free (content farms, deepfakes), while verification remains human, slow, and error-prone. The article urges product designers to treat the problem as a design challenge: add friction at creation (rate limits, signed origin records, proof-of-personhood), surface verifiable provenance and model uncertainty inline, and adopt portable standards (Content Credentials / C2PA) so provenance travels with media. It also notes regulatory pressure (EU AI Act labeling provisions enforceable August 2026) and recommends prioritizing explainability and shared proof infrastructure to reduce the asymmetry’s damage to trust across feeds and ad-supported products.

• Alberto Brandolini first expressed the asymmetry in 2013: refuting bullshit requires an order of magnitude more energy than producing it.
• NewsGuard counted 3,006 AI content-farm sites across sixteen languages as of March 2026 (up from 2,089 the previous October) and reported growth of 300–500 new sites per month; one analyst reportedly built a functioning content farm for about $100.
• Industry estimates cited in the article put deepfakes rising from roughly 500,000 in 2023 toward a projected 8 million in 2025.