Which Companies Now Use AI in Technical Interviews?
Updated 2026-11-04
The interview loop for software and AI engineers is changing faster than most prep guides can keep up with. A growing number of companies now hand candidates an AI assistant during technical rounds and evaluate how they use it. Others still prohibit AI entirely. This guide tracks what we know from official announcements, press reporting and candidate reports. No speculation.
The shift: from banning AI to requiring it
For most of the history of technical interviewing, outside assistance was cheating. That started to change in 2025. According to a Karat report based on over 600,000 interviews, almost two-thirds of companies still prohibit AI use in interviews. But the companies moving in the other direction tend to be large, high-profile employers, and the trend is accelerating.
As Cognition's head of people and operations, Emily Cohen, put it: banning AI in interviews is like "asking a kid to take a math test without a calculator."
Company-by-company breakdown
Google: Code comprehension with Gemini
Google is piloting an AI-assisted coding interview for junior and mid-level software engineering roles on select US teams. The pilot was first reported by Business Insider in May 2026, with October 2026 reporting indicating a broader rollout.
In a "code comprehension" round, candidates analyze an existing codebase (identifying bugs, improving performance and refining real-world code) with Gemini available as an assistant. An internal document says interviewers evaluate "AI fluency, including prompt engineering, output validation, and debugging skills."
Google's VP of recruiting, Brian Ong, said they're "always evolving our interview processes to ensure we're recruiting and hiring the best talent" and that the pilot is meant to "better reflect how teams work in the AI era."
Google has also said that AI fluency is required for all software engineering jobs, and that 75% of its new code is AI-generated.
Canva: AI-Assisted Coding replaces CS fundamentals
In June 2025, Canva replaced its Computer Science Fundamentals interview round with an AI-Assisted Coding round for backend, machine learning and frontend engineering candidates. Classic algorithm problems were swapped for open-ended, practical tasks. The company's example: replacing Conway's Game of Life with a system to manage aircraft takeoffs and landings.
Candidates choose their own AI tools (the company names GitHub Copilot, Cursor and Claude as examples). Interviewers assess whether candidates use AI effectively, understand vague requirements, make sound engineering decisions, find and fix bugs in AI-generated code and produce production-quality work.
Canva said nearly half of its front- and backend engineers already use AI daily, and that the old format (problems AI can solve in seconds) no longer measured real-world problem-solving. In the pilot, Canva noted that candidates with little AI experience struggled "not because they couldn't code, but because they didn't know how to work with AI or spot bad advice."
Meta: AI-enabled coding round (from October 2025)
Meta began piloting an "AI-enabled coding" round that replaces one of two onsite coding rounds. According to candidate reports, it is a 60-minute CoderPad session in a multi-file codebase with an assistant panel that answers questions in chat but cannot edit files directly. One problem covers a bug fix, a feature and an optimization.
Candidate reports describe four scoring areas: problem solving, code quality, verification and communication. Meta has confirmed it is "testing AI tools for applicants," but the details above come from candidate accounts, not an official guide.
CodeSignal: Cosmo AI assistant in assessments
In May 2025, CodeSignal launched AI-assisted coding assessments with its built-in AI assistant, Cosmo. Organizations can configure Cosmo in two modes: a full co-pilot that actively helps with complex challenges, or a guided mode that offers lighter help like syntax fixes and documentation. Hiring teams get a full transcript of every candidate-AI interaction and a session replay showing how candidates used the assistant.
Karat: NextGen interviews
In December 2025, Karat launched NextGen interviews: a human-led, AI-enabled format where candidates work on complex multi-file projects with a built-in AI assistant while live interviewers assess reasoning, trade-offs and judgment. Karat co-founder Jeffrey Spector said: "Most companies are still hiring based on a pre-LLM rubric."
Zapier: AI fluency required for every hire
Zapier went further than interview format: since May 2025, AI fluency is a hiring requirement for every new hire across all roles, not just engineering. The company's V2 rubric (March 2026) defines four levels:
- Unacceptable: Used only as a light assist with no measurable impact
- Capable (minimum bar): Uses AI to operate at a meaningfully higher level, with repeatable systems and measurable impact
- Adoptive: Builds systems that improve work beyond the individual
- Transformative: Redesigns how work happens, not just how fast it gets done
Candidates are watched using AI live, and Zapier's Global Head of Talent, Tracy St.Dic, wrote that "with AI, you can delegate the work, but not the accountability." Zapier reports 100% AI adoption across teams.
Other companies allowing AI
- Shopify and Rippling explicitly allow candidates to bring AI copilots into live coding sessions.
- LinkedIn has an AI-assisted round that runs alongside a traditional coding round, per candidate reports.
Companies that still prohibit AI
- Anthropic, Microsoft and OpenAI say AI is off unless explicitly permitted (OpenAI notes it varies by interview).
- Amazon's internal recruiter guidelines (reported February 2025) prohibit generative AI unless explicitly permitted.
- Datadog (May 2026) bans AI in live coding and technical assessments, except in interviews specifically designed for AI-assisted coding, which candidates must be told about in advance.
What these interviews have in common
Despite the different formats, the AI-assisted rounds share a common structure:
- You start from existing code, not an empty file
- Something is wrong or missing: a bug, a failing case, a slow path
- An AI assistant is available, and how you use it is part of the evaluation
- You explain your reasoning throughout. The thinking is what gets scored, not just the output
Companies consistently report that the strongest candidates treat the AI as a tool they supervise, not an oracle they obey.
What to do before your interview
Ask your recruiter in writing which interview loop applies and which tools are allowed. Formats vary by team even within the same company. Don't assume.
If you want to practice the format (reading unfamiliar AI code, finding the bug and explaining the fix with an AI assistant available), that's what GenAIDojo's Code Comprehension track is built for. You can try a problem before you log in.
Sources
- Google Is Testing a Transformative New Interview Rule, Entrepreneur
- Google Changed the Coding Interview: AI Is Now the Co-Pilot, ByteHerald
- Canva Requires Engineers to Use AI in Interviews, Scroll.media
- AI-Assisted Coding Interviews: What to Expect and How to Prepare, GreatFrontEnd
- CodeSignal Launches AI-Assisted Coding Assessments and Interviews, CodeSignal
- Karat Launches NextGen Interviews, Karat
- Raising the AI Fluency Bar in Hiring, Zapier
- LinkedIn: AI Engineering, Prompting & Model Tuning Are the Fastest-Growing Skills in 2026, Interview Query