← Back to the wire

Worrying about bad AI code is worrying about the wrong problem

AnnouncementProductSep 13, 2026

Matt Horn argues that concerns over AI-generated code quality distract from more serious AI risks. He cites researchers departing major labs: Jacob Coxon resigned on September 8, saying labs are racing toward self-improving superintelligence, and Alex Turner left DeepMind with similar concerns; Anthropic's alignment lead put extinction odds above 10% this decade. Horn also points to reported agent incidents and math breakthroughs, arguing AI systems cannot be dismissed as incapable while being treated as dangerous.

Receipt № 19011 source · awaiting confirmation ◐

Evidence

1source· awaiting independent confirmation

No score is assigned. Sources and their independence are shown in the citation chain below.

Citation chain · 1 source

01highPRIMARY
OpenAICompanyAnthropicCompanyDeepMindCompanyJacob CoxonPersonMatt HornPersonAlex TurnerPerson
Canonical: https://matthorn.io/posts/2026-09-13-bad-ai/