Secure your remote interviews with real-time AI fraud detection.
Grade: B — Score: 83/100
Sherlock AI employs a multimodal adversarial machine learning approach to identify interview fraud, integrating signals from device activity, audio environments, and candidate behavior into a unified classifier. This advanced technology allows for the detection of subtle cheating tactics that traditional methods may miss.
The platform seamlessly integrates with existing workflows by connecting to calendars and automatically generating secure meeting links for interviews. During the interview, Sherlock provides real-time alerts for suspicious activities, allowing interviewers to focus on evaluating candidates' skills.
As remote interviews become more prevalent, the risks of AI-assisted cheating increase. Sherlock AI mitigates these risks by continuously refining its detection capabilities, ensuring high accuracy and reliability in maintaining the integrity of the hiring process.
Fraud Detection Starter: $100/month monthly; $80/month billed annually
Fraud Detection Growth: $1,500/month monthly; $1,200/month billed annually
Fraud Detection + AI Fluency Starter: $120/month monthly; $96/month billed annually
Fraud Detection + AI Fluency Growth: $1,800/month monthly; $1,440/month billed annually
Fraud Detection + AI Fluency + Note Taking Starter: $200/month monthly; $160/month billed annually
Fraud Detection + AI Fluency + Note Taking Growth: $3,000/month monthly; $2,400/month billed annually
Enterprise: Custom quote
Consider switching to HireVue: HireVue also offers video interviewing solutions with AI capabilities but lacks the same level of real-time fraud detection.
Sherlock AI is designed to flag possible unauthorized AI assistance, and its help FAQ specifically names ChatGPT. It combines device, audio and behavioral signals during live interviews. This does not establish that every assistant or evasion method will be detected; use alerts as supporting information for interviewer review.
Sherlock AI advertises detection of deepfakes and proxy candidates, with a dedicated real-time deepfake mode that consumes five credits per scheduled session. The reviewed materials do not establish dedicated detection coverage for AI glasses. Ask Sherlock to demonstrate the wearable or off-device scenarios relevant to your interviews.
Sherlock AI claims that adversarial retraining increased detection accuracy from about 85% to more than 97%, and its general help FAQ claims a false-positive rate below 2%. These are vendor-reported figures, not independently validated results. The reviewed pages do not provide the test-set composition, sample size, thresholds or complete evaluation method needed to assess their applicability. Request validation details for your interview conditions and review flagged sessions in context.
Installation depends on the selected Sherlock AI mode. Detect Only and Realtime Deepfake Detection + Detect do not require candidate installation, while Prevent + Detect requires a desktop helper application. Confirm the mode and required permissions before sending candidate instructions.
Sherlock AI says video, audio or screen recording occurs when the interviewer enables it, with a visible recording indicator. Its general FAQ describes AWS-hosted data, while a separate storage FAQ says data stays on the user's system unless explicitly shared; the reviewed documentation does not fully map those statements to data categories and modes. Account retention settings offer 7 days, 14 days or retention until manual deletion for data created after the setting is saved. Confirm the recording configuration, hosting region and handling of optional biometric identity checks before deployment.
Sherlock AI describes its outputs as advisory and states that it does not make final hiring or employment decisions. Employers remain responsible for interpreting the findings and deciding how to proceed. Its integrity and AI Fluency scores should not be confused with either a final hiring decision or proof that no candidate scoring takes place.
Sherlock AI lists Google, Apple and Outlook calendars, and its help center documents Zoom and Microsoft Teams workflows. Its general FAQ names Greenhouse, Lever, Workday and iCIMS integrations through APIs, as well as Calendly scheduling integration. Confirm connector availability, data flows and plan eligibility; a documented API connection is not necessarily a ready-made connector included in every plan.
Yes. Sherlock AI offers AI Fluency Observation to evaluate how effectively candidates use permitted AI tools. Its packages distinguish Fraud Detection, Fraud + AI Fluency, and Fraud + AI Fluency + Note Taking. Define the permitted tools and use cases before the interview so candidates understand the rules.
Sherlock AI's Privacy Policy says it does not train AI models on personal data. The DPA specifies that Customer Personal Data is not used to train or fine-tune general-purpose models for other customers unless the customer explicitly opts in in writing. It also permits de-identified or aggregated data for analytics and service improvement, subject to safeguards against identification.
Sherlock AI focuses on integrity monitoring for live, human-led interviews, with Fraud Detection Starter listed at $100 per month for 10 interviews on monthly billing. Interviewer.AI Professional lists $399 per month with 120 interview credits, fraud checks and conversational AI avatar interviews alongside one-way video screening. Sherlock is a closer fit when you want to monitor interviews your team conducts; Interviewer.AI offers a broader automated screening workflow. Interview credits are not directly interchangeable between the products.
How AI agents (ChatGPT, Perplexity, Claude, others) read this review page in the past 7 days. Updated weekly. View Sherlock AI AI Visibility Report.