Terms of Service

1. Agreement and eligibility

These Terms are a binding agreement between First Round AI and each Customer, account user, candidate or guest using the service. You must be at least 18 or the age of legal majority and authorized to bind any organization you represent. By clicking Agree, typing a legal name, starting a disclosed guest interview, or using the service after notice, you electronically sign under E-SIGN and UETA and agree that server records may prove assent. If you decline required terms, do not use the service.

2. Service and accounts

The service provides ElevenLabs synthetic-voice interviews, transcription, multilingual support, rubrics, AI summaries/scores, applications, scheduling, alerts, workforce tools, coaching, assistants and related recruiting workflows. Keep credentials secure, provide accurate information and promptly report unauthorized access. Customers administer their users and are responsible for workspace activity, lawful configurations, notices, consents and retention choices.

3. Fees, subscriptions and purchases

Subscriptions renew at the disclosed cadence until canceled. Fees, taxes and usage charges are processed by Base44 Payments or another third-party checkout; we do not store full card numbers. Except where law requires otherwise, fees are non-refundable, including unused periods. Price changes apply after notice to a future renewal. Promotional codes have stated limits and may be withdrawn for abuse. Fraudulent chargebacks, evasion or account cycling may result in suspension and collection costs. One-time coaching or generated deliverables are final and non-refundable once delivered.

Marketing SMS consent is optional and not a condition of purchase; message/data rates and frequency disclosures apply, and STOP/HELP controls must be honored. Transactional verification or interview messages may be sent as requested.

4. Ownership and licenses

First Round owns the platform, models, software, synthetic interviewer personas and authorized voice designs. Customers retain ownership of their content and grant First Round a limited worldwide license to host, process, transmit and display it to provide, secure and improve the service. We may improve systems using aggregated or de-identified data that does not reasonably identify a person or Customer. Feedback may be used without restriction. No license permits extracting models, recording or sampling the AI voice, cloning a persona, or creating deepfakes or impersonations.

5. Customer content and compliance

Customers warrant they have rights and lawful bases for candidate data, job content, recordings, custom voices and instructions. Customers are the employer/controller/deployer and must comply with Title VII, ADA, ADEA, GINA, BIPA, Illinois AIVIA, NYC Local Law 144, the EU AI Act, Colorado AI Act and other applicable employment, biometric, accessibility, pay-transparency, recording and AI laws. Customers must provide accommodations and alternative interview channels where required.

6. AI outputs and hiring decisions

AI outputs are probabilistic, experimental, may be incomplete or biased, and are provided as is. They are not legal, employment, medical or professional advice and may not be the sole or determinative basis for employment action.

Meaningful human review is a binding condition of use. A qualified human decision-maker is a person trained for the role, authorized to change the outcome, familiar with the job criteria and applicable anti-discrimination/accommodation duties, and able to inspect source responses, transcript context, limitations and contrary evidence. Meaningful review requires independent judgment, documented consideration of relevant context, correction of errors, accommodation handling, and a genuine ability to override the AI. Automatic approval, rubber-stamping, reviewing only a score, or treating an AI recommendation as presumptively correct is not meaningful review. No solely automated hiring decisions are permitted.

Customers must evaluate disparate impact, provide legally required notices/consent and bias audits, monitor outcomes, and never use visual, sentiment, attention, integrity or language signals to infer protected traits. We do not estimate apparent ethnicity, gender or age. Automated translation can err; high-stakes multilingual outcomes require review in the relevant language.

7. Feature-specific terms

Guest demo consent is logged with version, timestamp, IP/device evidence and displayed language under E-SIGN/UETA. Job-listing output is a draft; pay ranges, classification and transparency compliance remain the employer's responsibility. In application flows, a typed legal name may constitute an electronic signature and applicant consent is logged server-side. Customers control scheduling, offers, alerts and workforce rules and must observe wage-and-hour laws.

Realtime speech-to-text is the official transcript and is retained/scored like batch transcription. Program, fellowship or data-training terms may require a separate agreement; unless expressly stated, program work may be independent-contractor work-made-for-hire, and sports reasoning is not betting advice. Cohort uploads may pass an NSFW safety filter but Customer review remains required. Field/canvassing features require lawful location and recording consent.

Users must configure AI assistants and agents with least privilege. Outreach must comply with CAN-SPAM, TCPA, CASL and ePrivacy, including consent, identification, quiet hours and opt-out suppression. Custom voice users warrant documented rights and consent and may not imitate a person deceptively. Metering, audit logs, rate limits, bias controls, trust fields and server evidence may not be bypassed or altered.

8. Acceptable use and candidate treatment

The Acceptable Use Policy is incorporated. No discrimination, harassment, unlawful surveillance, security abuse, consent falsification, scraping, model extraction or deceptive impersonation is allowed. Candidates must receive clear notice of synthetic voice, recording/video, evaluation purpose, data controller and available alternatives.

9. Third parties; suspension

Third-party services have their own terms and may change or fail. We may suspend access for legal, security, payment or material policy violations and may preserve evidence. Either party may terminate as allowed by an order form; accrued duties and provisions intended to survive remain effective.

10. Disclaimers, liability and indemnity

THE SERVICE IS PROVIDED AS IS AND AS AVAILABLE, WITHOUT IMPLIED WARRANTIES OF MERCHANTABILITY, FITNESS, NON-INFRINGEMENT, ACCURACY OR UNINTERRUPTED OPERATION. No SLA applies unless signed. To the maximum extent allowed, First Round is not liable for indirect, special, consequential, exemplary or lost-profit/data damages. Aggregate liability is capped at fees paid for the affected service in the preceding 12 months, and claims must be brought within one year. These limits do not exclude liability that law makes non-waivable.

Customers indemnify First Round from claims arising from employment decisions, discrimination, biometric/recording consent, Customer content, unlawful outreach, custom voices, or Customer violation of AI/employment law. First Round controls defense subject to reasonable cooperation.

11. Disputes and general terms

Before filing, contact Support and allow 30 days for informal resolution. Except small-claims matters or injunctive IP/security relief, disputes are resolved by binding individual AAA arbitration: Consumer Rules for consumers and Commercial Rules for businesses. The parties waive class, collective and representative proceedings and jury trial. Consumers may opt out by written notice through Support within 30 days of first acceptance, identifying the account and request. Applicable non-waivable local rights remain.

Delaware law governs without conflicts rules; permitted court matters lie in Delaware state or federal courts. Neither party may assign without consent except in a merger or asset transfer. Notices may be electronic. Invalid terms are severed; waiver must be written; these Terms and incorporated documents are the entire agreement. Material changes are published in advance; the prior version governs until the effective date and re-confirmation may be required. See Privacy Policy §1.4.

Server records, electronic assent and source of truth

The authoritative record of assent is created and verified on the server. The Service hashes the exact document text with SHA-256 and signs the acceptance envelope with HMAC-SHA-256 using a secret unavailable to browser code. Browser storage, screenshots, client-provided flags, or altered requests cannot establish current acceptance. A valid acceptance must verify cryptographically, identify the user, policy and version, and match the currently published version. Publishing a new effective version makes an older acceptance outdated and requires renewed assent before gated account areas can be used. Hashing is an integrity control rather than encryption; sensitive fields are separately encrypted where described in the Privacy Policy.

Last revised September 15, 2026 (Version 1.1). Previous versions are available in the policy archive.

2026 Platform Protections Addendum

1. AI output limitations

AI-generated or automated outputs may be incomplete, inaccurate, delayed, or unsuitable for a particular purpose. Users must apply qualified human review before relying on material decisions.

2. No professional advice

The service does not provide legal, medical, financial, employment, safety, engineering, or other regulated professional advice.

3. Feature availability and discretion

Plans, models, integrations, credits, limits, prices, and features may change. FirstRound AI may limit, pause, replace, or disable a feature at its sole discretion for safety, compliance, provider availability, abuse prevention, or product operations.

4. Credits, billing, and taxes

AI credits fund recorded provider costs, are not cash, and are not transferable unless expressly stated. Taxes may apply. Recurring credits do not roll over unless checkout terms expressly say otherwise.

5. Third-party services

Connected providers, models, payment processors, data sources, and external links operate under their own terms and may change or become unavailable without notice.

6. Account and credential security

Users are responsible for safeguarding accounts, links, tokens, devices, and credentials and for promptly reporting suspected unauthorized use.

7. Content rights and permissions

Users must have the rights and permissions needed for uploaded, generated, recorded, or shared content and must not submit unnecessary secrets or regulated personal data.

8. Demo, alpha, and beta services

Demo, alpha, beta, preview, and experimental services are provided for evaluation, may reset or contain simulated data, and must not be treated as production guarantees.

9. Abuse prevention and enforcement

FirstRound AI may investigate, rate-limit, suspend, remove, preserve, or disclose activity as reasonably required to prevent fraud, abuse, harm, infringement, or legal violations.

10. Policy changes and continued use

Material changes are published as a new version and may require renewed acceptance. Continued use after the effective date constitutes acceptance where permitted by law.