Recording & AI disclosure

This policy is part of, and incorporated into, the avoClerk Terms of Service. It describes how avoClerk discloses AI use and call recording to callers, the per-state consent rules that apply, and the responsibilities of the firm (the customer) versus avoClerk (the vendor).

The two non-negotiable disclosures

Every avoClerk call begins with two disclosures baked into the greeting:

1. AI identity disclosure — the caller is told they are speaking with an AI assistant. This runs on every call, regardless of the caller’s state. It is required by California AB 2905 ($500 per undisclosed AI call, effective January 1, 2025), California SB-243 (AI bot disclosure when asked), Florida SB-998, the FTC Act § 5 prohibition on deceptive practices, and analogous emerging laws.
2. Recording disclosure — when the firm has configured the consent mode to “two-party”, the caller is told the call is recorded before any substantive conversation. When the consent mode is “one-party”, the recording disclosure is omitted.

The firm may not disable AI identity disclosure. The firm may, by configuration, run with recording entirely off.

Illinois — declined at the network edge

avoClerk does not currently serve callers with Illinois area codes (217, 224, 309, 312, 331, 447, 618, 630, 708, 773, 779, 815, 847, 872). Inbound calls from those area codes are declined with a brief decline message at the inbound webhook before any speech-to-text or large-language-model component runs and before any audio is recorded or stored. The reason is the Illinois Biometric Information Privacy Act (740 ILCS 14), which combines per-violation statutory damages, a private right of action, and class-action machinery in a way no other US state currently does, and whose application to modern speech-to-text and large-language-model pipelines is unsettled.

The firm may not configure or use the Service in a manner intended to circumvent this restriction.

Two-party / all-party consent states

The federal Wiretap Act (18 U.S.C. § 2510 et seq.) is one-party consent. State law overrides where stricter. The following table reflects our understanding as of the last-updated date.

One-party consent applies in every state not listed above. The caller’s state controls. AI disclosure obligations may apply regardless of one-party / two-party status.

Per-firm consent mode

The firm sets one global consent mode in the dashboard:

• two_party (default for safety) — the recording disclosure runs in the greeting. We strongly recommend this for any firm whose callers may include CA, CT, FL, MD, MA, MI, MT, NV, NH, PA, or WA residents.
• one_party — no recording disclosure in the greeting. The firm represents that its expected caller pool is in one-party-consent states only.
• no_recording — no recording at all. The Service still transcribes for intake purposes but does not retain audio. Real-time speech-to-text may itself constitute interception under stricter state statutes, so this is not a complete shield.

Canonical greetings

two_party (default)

“Hi, you’ve reached [Firm Name]. You’re speaking with an AI assistant, and this call is being recorded to help with your intake. If that works for you, please go ahead — how can I help today?”

one_party

“Hi, you’ve reached [Firm Name]. You’re speaking with an AI assistant. How can I help today?”

no_recording

“Hi, you’ve reached [Firm Name]. You’re speaking with an AI assistant, and we’re not recording this call. How can I help today?”

Caller declines consent (two-party mode)

If, after the recording disclosure in two_party mode, the caller indicates they do not consent, the Service immediately stops audio recording, persists no recording for that call, and continues only in transcript-only mode if the caller continues the conversation. The Service does not interpret silence as consent.

Mid-call AI re-disclosure

If the caller asks “are you a real person?” or any equivalent, the Service affirmatively discloses AI identity. This is implemented as a non-overrideable rule in the system prompt.

No legal advice

If a caller appears to ask for legal advice, the Service responds with a configured disclaimer and offers to take a message for the firm’s licensed attorneys to follow up. This safeguard is non-removable.

Audit trail

For every call, the Service captures: timestamp, consent mode in effect, greeting variant played, AI-disclosure event, recording-disclosure event (where applicable), consent-capture signal, mid-call AI-re-disclosure events, speaker-change events triggering re-disclosure, retention window, and whether audio was retained. The firm can export the trail per call from the dashboard.

Responsibilities

The firm picks the consent mode appropriate for its caller pool, reviews and accepts the dashboard attestation, trains staff on handling caller questions about recording or AI use, maintains its own copy of consent audit logs, and supervises the Service as a non-lawyer assistant under ABA Model Rule 5.3.

avoClerk runs the disclosures, captures the audit trail, provides the configuration controls, and updates this policy on at least thirty (30) days’ notice for changes affecting firm-side configuration.

Avo LLC, 30 N Gould Street, STE R, Sheridan, WY 82801.