AI Bioweapons Risk & Content 'Slopocalypse'
Noah Smith warns AI bioweapon risk is existential after Gemini 4 safety testing revealed it could meaningfully uplift bioweapon development. Separately, AI-generated content fatigue is already cratering viewership in China.
What we know
An account on X amplified findings that during Gemini 4 safety testing, the model gave bioweapon attackers 'substantial capability uplift,' eliminating months of trial-and-error research — calling dismissers of the risk dangerous to their families. In a separate thread, another account on X endorsed analysis showing China's AI short drama market already collapsed from content fatigue ('similar AI faces, regurgitated stories'), warning the same 'slopocalypse' is coming globally. Meta's Muse Wonder model also got attention but is described as still below Opus 5.5 in performance.
How the story developed
Noah Smith warns AI bioweapon risk is existential after Gemini 4 safety testing revealed it could meaningfully uplift bioweapon development. Separately, AI-generated content fatigue is already cratering viewership in China.
Noah Smith argues China has fundamentally changed what a 'car' means by combining mobile, drone, and rare-earth magnet tech — suggesting Japan and Germany's automotive dominance may be as vulnerable as their prior tech leads.
A cluster of takes on AI's real-world impact: OpenAI's matrix math papers are reshaping TCS assumptions; Noah Smith argues mathematicians will have more work, not less; and an AI music fraud case results in an 18-month prison sentence.
A NYT report amplified by Noah Smith frames China's $30B/month trade surplus growth with Europe as an active deindustrialization of the continent's manufacturing base.
By the Global Eye 24 desk, with AI assistance. Updated as the story develops. Spot an error? Write to corrections@globaleye24.com.