Essay: AI-era information intermediaries pose systemic danger even in careful hands

sebkrier · x · 2026-08-27

A new essay argues that new intermediaries emerging in the age of AI hold systemically dangerous power even in careful hands, and that keeping their power checkable and contestable is how we can reap AI's benefits without surrendering pluralism, privacy, or safety.

The author draws historical parallels: the Mythos export controls echo the Crypto Wars, and open-weights letters mirror the Free Software movement. AI's information dynamics are uniquely risky — it performs better with more data and digests data extremely well, a dangerous flywheel — with recurring worries around AI-enabled surveillance and influence campaigns.

Original post →

More from AGI Musings

AGI Musings channel →