There is a structural irony baked into the current AI coding moment: the same systems generating unprecedented volumes of software are also generating unprecedented volumes of unverifiable software. Qodo's $70M raise is a direct bet that this gap between generation and verification is not a bug to be patched but a permanent feature of the AI development cycle. Someone has to audit the machine.
The BeSafe Benchmark and AI Behavioral Risk
The timing is not coincidental. This week's BeSafe-Bench paper from arXiv introduced a framework specifically for benchmarking the behavioral safety risks of AI agents operating in functional environments. The paper's core finding: current large multimodal models exhibit safety failures not in adversarial edge cases, but in ordinary, situated task completion. This is the same problem Qodo is selling against, just one layer up the stack. When AI agents write code for functional systems, including buildings, airports (see this week's airport knowledge engineering paper), and power grids, the verification gap becomes a liability gap. The AIRA_2 paper on AI research agents identified three structural bottlenecks in autonomous AI systems, and notably, verification of outputs was listed as the hardest to solve programmatically. Qodo is commercializing the hardest problem.
Code Quality as the Next Infrastructure Bet
Adobe Illustrator's Turntable feature launch this week is a softer version of the same tension: AI-assisted creative tools require new verification loops because the outputs no longer map cleanly onto the creator's intent. In code, that gap is a security vulnerability. In design, it is a copyright question. In both cases, the human-in-the-loop is being repositioned from producer to auditor. A 2025 paper in IEEE Transactions on Software Engineering by Zheng et al. found that AI-generated code contained security vulnerabilities at a rate 40% higher than human-written equivalents when used without verification tooling. That number is Qodo's pitch deck. TurboFund's list of 25 AI seed investors shows heavy clustering around safety and verification infrastructure in recent check activity, suggesting Qodo's $70M is the first of several large rounds in this category.