Can You Use AI in Coding Interviews? What Changed in 2026

In 2026, using AI in coding interviews now depends on the specific round of the interview process, as evidenced by recent changes in company policies. AI-assisted coding screens are now standard for backend and frontend roles at Canva, following their June 2025 announcement. However, Anthropic has maintained that AI should generally not be used in live interviews or take-home assignments unless explicitly permitted, as seen in their July 2025 guidance updated in January 2026.

Microsoft's technical interview guidelines remain consistent, covering problem-solving, coding, testing, algorithms, data structures, and complexity analysis, without specifying AI usage. Employers should be questioned directly about AI usage for each round of the interview process, as CodeSignal's proctored assessments monitor for allowed tools such as AI chat and code generation, but do not grant permission by default.

Evaluation criteria for AI-allowed rounds vary by company. Canva now assesses engineering problem-solving, technical depth, ownership, and AI collaboration, while Anthropic's hiring scorecards from October 2026 show a significant increase in comments on autonomy and judgment, indicating a shift in what employers value in AI-assisted rounds.

In practice, candidates should clarify requirements, review every generated change, test the assistant's behavior, and defend their decisions if the assistant produces different results.

For take-homes, AI permission is not automatic, and candidates should verify whether external AI services can be used and how to submit their work. A practice plan for both unassisted and assisted interviews is provided, highlighting the importance of clarifying assumptions, writing tests, and understanding the contributions of AI versus the candidate's own work.