Engineers Warn 75% AI-Written Code Can Spawn Dangerous 'Vibe Slop'
Updated
Updated · InfoWorld · May 25
Engineers Warn 75% AI-Written Code Can Spawn Dangerous 'Vibe Slop'
2 articles · Updated · InfoWorld · May 25
Mario Zechner and Armin Ronacher, engineers behind parts of the OpenClaw AI agent, warned AI coding tools are flooding software with low-quality "vibe slop" that can slip buggy, dangerous code into production.
Their concern is not obvious failure but plausible output that lets users skip design, testing, ownership and system understanding—the parts of software work that catch errors before release.
GitHub has already weighed tighter pull-request controls after maintainers said low-quality, often AI-generated submissions were overwhelming open-source projects and making review costs spike.
Google says 75% of its new code is AI-generated and then reviewed by engineers, underscoring the article's point that AI helps most when experienced humans apply judgment, guardrails and accountability.
As AI floods open source with 'vibe slop,' is it destroying the very communities it was meant to empower?
With AI writing most new code, are we training engineers to be expert critics or just glorified rubber-stampers?
"Vibe Slop Crisis: Over 30% of AI-Generated Code Contains Critical Vulnerabilities—Security, Economic, and Regulatory Impacts in 2026"
Overview
The report highlights an urgent crisis in the digital world called "Vibe Slop," where AI-generated code is created rapidly without enough security checks or human review. This has led to a sharp rise in software vulnerabilities and maintenance problems, putting heavy pressure on development teams. A key example is the Moltbook breach in early 2026, where a social network built quickly with AI suffered a major data leak due to poor security controls. As more platforms face similar risks, the industry is scrambling to adopt stronger defenses and rethink how AI-driven development should be managed to protect users and maintain trust.