FlashCrafter

Guide

AI Tools Security and Privacy Checklist for Local Service Businesses: Customer Data, Call Recording and Text Consent

What an owner-operator should check before an AI receptionist, chatbot, call recorder or texting tool touches customer data, with the US consent rules, costs and penalty figures that apply.

Owners of independent local service businesses in the United States (home services, trades, clinics, salons and similar) who use or are considering AI receptionists, chatbots, call recording, AI note-takers or business texting.

AI Tools Security and Privacy Checklist for Local Service Businesses: Customer Data, Call Recording and Text Consent
AI Tools Security and Privacy Checklist for Local Service Businesses: Customer Data, Call Recording and Text Consent
Key facts
9
Steps
13
Pitfalls covered
8
Research captured
2026-10-01

An AI phone agent, a website chatbot or a texting tool does more than save time. It listens to your customers, stores what they say, and often sends it to a third-party vendor. That makes you responsible for three things at once: how the vendor handles the data, whether callers and texters agreed to be recorded or contacted, and whether you can prove it later.

Most published checklists are written for call centers or enterprise IT teams. This guide is written for an owner who has a phone, a Google Business Profile, maybe Google Ads or Local Services Ads, and a handful of software subscriptions. It covers the rules that matter most in practice: call-recording consent by state, text-message consent under the TCPA, the 10DLC registration that business texting requires, the data-use clause in your AI vendor's contract, and the call recording that Google Ads can switch on for you by default.

Two cautions apply throughout. First, this is a practical checklist, not legal advice. Several figures below come from law-firm and trade summaries rather than the statute or court text, and each place where that matters is flagged. Second, where published counts disagree (for example, the number of all-party-consent states), the disagreement is stated rather than smoothed over. Information is current as of October 1, 2026.

Why local service businesses should care

Breach and fraud statistics are mostly gathered across organizations of all sizes, but several point directly at small firms.

Verizon's 2025 Data Breach Investigations Report, published April 23, 2025, analyzed more than 22,000 security incidents, of which 12,195 were confirmed breaches. Third-party involvement in breaches doubled to 30%. Ransomware appeared in 44% of breaches overall and in 88% of breaches involving small and medium businesses. The median ransom paid was $115,000, and 64% of victims did not pay. For initial access, credential abuse accounted for 22% of breaches and vulnerability exploitation for 20%. The 88% figure comes from Verizon's press release; the report PDF itself was not parsed, though Security Magazine repeats the same numbers.

IBM and Ponemon's Cost of a Data Breach report is the other widely quoted source. IBM's 2026 report landing page shows a global average of $4.99M, a 12% increase over the prior year and a record high, a 56% rise in AI-driven attacks, and $1.93M saved by extensive use of security AI and automation. Secondary coverage of the 2026 report puts the US average at $11.5M (Security Boulevard, August 2026) and says 92% of organizations with AI-related incidents lacked adequate AI access controls (eSecurity Planet). Those two figures were not on IBM's page as retrieved. For the 2025 report, secondary sources say high levels of shadow AI added $670K to average breach cost and that 97% of organizations with AI-related incidents lacked AI access controls. One page carrying these figures also carries a 2026 date, so check which report year a figure belongs to before you repeat it. IBM's figures describe organizations in general; the page does not break out small businesses, and no primary-source average breach cost for a local service business under 50 employees has been found.

The FBI's Internet Crime Complaint Center (IC3) 2025 report, released April 7, 2026, counted 1,008,597 complaints and $20.877B in reported losses, up 26% on 2024. Business email compromise accounted for about $3.05B. The report introduced 'AI-related' as a formal descriptor, with more than 22,000 complaints and nearly $900M in losses. These figures come from secondary summaries (SpyCloud and Hoodline) of the primary report PDF at ic3.gov.

Litigation over calls and texts is also rising. Using WebRecon data reported by TCPAWorld, Q1 2025 TCPA class action filings were 507, up 112% from 239 in Q1 2024, and September 2025 class action filings spiked 283%. About 80% of TCPA suits are class actions. These are blog-level tallies, not court data.

Published breach and fraud figures at a glance

ReportDateFigureCaveat

Verizon 2025 Data Breach Investigations Report

April 23, 2025

22,000+ incidents; 12,195 confirmed breaches; third-party involvement 30% (doubled); ransomware in 44% of breaches and 88% of SMB breaches; median ransom $115,000; 64% did not pay; credential abuse 22%, vulnerability exploitation 20%

88% SMB figure is from the press release, not the report PDF

IBM Cost of a Data Breach 2026

2026

Global average $4.99M (up 12%); 56% rise in AI-driven attacks; $1.93M saved by extensive security AI and automation

Organizations in general; no small-business breakdown on the page

IBM 2026, secondary coverage

August 2026

US average $11.5M; 92% of organizations with AI-related incidents lacked adequate AI access controls

Not on IBM's page as retrieved

IBM 2025, secondary coverage

2025

Shadow AI added $670K to average breach cost; 97% of organizations with AI-related incidents lacked AI access controls

Secondary sources only; one page carrying them also bears a 2026 date

FBI IC3 2025 report

April 7, 2026

1,008,597 complaints; $20.877B losses (up 26%); business email compromise about $3.05B; AI-related: 22,000+ complaints, nearly $900M

From secondary summaries; primary PDF is at ic3.gov

TCPA class actions (WebRecon via TCPAWorld)

Q1 2025; Sept 2025

507 in Q1 2025 vs 239 in Q1 2024 (up 112%); September 2025 up 283%; about 80% of TCPA suits are class actions

Blog-level tallies, not court data

What AI receptionists and answering services cost, and why retention is a priced feature

Security features are often tied to plan tier, so price and privacy are linked. The vendor pricing pages below were checked on October 1, 2026.

Smith.ai's page shows only live-receptionist-style tiers: Starter at $300 a month (30 calls, $11.50 overage), Basic at $810 (90 calls), Pro at $2,100 (300 calls) and Enterprise at custom pricing. Call recording and transcription is an add-on at $0.25 per call. Secondary sites list a separate Smith.ai AI plan at $150 a month (about $2 per call) or $97.50 a month, but that AI-only pricing does not appear on Smith.ai's own page, so confirm it with the vendor.

Goodcall prices by agent and by unique customers: Starter at $79 a month per agent (100 unique customers, 7 days of call details), Growth at $129 (250 customers, 30 days) and Scale at $249 (500 customers, unlimited call-detail retention). Overage is $0.50 per customer, and annual billing is 15% off. Because the retention window changes with the plan, how long your customers' call details are kept is something you pay for. A shorter window means less data exposed in a breach but less history for you.

Ruby, which uses live receptionists, lists Startup at 50 minutes for $250, 100 minutes for $395, 200 minutes for $720 and 500 minutes for $1,725.

AI receptionist and answering service pricing (vendor pages, seen October 1, 2026)

VendorPlans and pricesSource

Smith.ai

Starter $300/mo (30 calls, $11.50 overage); Basic $810 (90 calls); Pro $2,100 (300 calls); Enterprise custom. Call recording and transcription $0.25/call.

smith.ai/pricing/receptionists

Goodcall

Starter $79/mo per agent (100 unique customers, 7 days of call details); Growth $129 (250, 30 days); Scale $249 (500, unlimited). Overage $0.50 per customer. Annual billing 15% off.

goodcall.com/pricing

Ruby (live receptionists)

Startup 50 min $250; 100 min $395; 200 min $720; 500 min $1,725.

ruby.com/pricing

Call recording consent: federal law, state law and the all-party list

Federal law requires the consent of only one party to a call. Several states require everyone on the call to agree, and a business serving customers across state lines can hit those rules without ever leaving home.

Published counts of all-party states differ. Viirtue lists 11 (California, Delaware, Florida, Illinois, Maryland, Massachusetts, Montana, Nevada, New Hampshire, Pennsylvania and Washington). CommLaw Group lists the same set plus Connecticut. Frontdesk and AInora count 12, and Dialzara counts 12 all-party states and 38 one-party states. Recordinglaw.com counts 13, adding Connecticut and Oregon, with caveats about whether the rule applies to phone-only or in-person-only recording. The disagreement concerns Connecticut, Michigan, Oregon, Vermont and Delaware. No primary 50-state source was available (Justia returned an access error and the RCFP index did not list details), so the counts here are secondary.

Practical rule: recordinglaw.com says that if either party is in an all-party state, the safest approach is to get everyone's permission. It also says California law applies to calls to or from California regardless of the caller's location. For a service business that takes calls from customers who may be in other states, an announcement at the start of every recorded call, whether from a human or an AI agent, avoids having to guess.

How published lists count all-party-consent states

SourceCountNotes

Viirtue

11 states plus mixed cases

CA, DE, FL, IL, MD, MA, MT, NV, NH, PA, WA

CommLaw Group

12

Same set as Viirtue plus Connecticut

Frontdesk; AInora

12

Counts as stated; no state-by-state detail in the guide

Dialzara

12 (and 38 one-party)

Counts as stated

Recordinglaw.com

13

Adds Connecticut and Oregon, with caveats for phone-only or in-person-only scope

AI voice and wiretap lawsuits: why your vendor's data-use clause matters

Two cases show how a vendor's behavior can turn a recorded call into a legal claim. The checkable item in both is the same: does the vendor keep and use conversations for its own purposes, such as training its models?

In Ambriz v. Google (N.D. Cal., February 10, 2025), the court denied dismissal on the theory that an AI call-center vendor with the capability to use call data for its own purposes can be a third-party eavesdropper under the California Invasion of Privacy Act (Goodwin and ZwillGen summarize the ruling).

In In re Otter.AI Privacy Litigation, No. 5:25-cv-06911-EKL (N.D. Cal.), Judge Eumi Lee ruled on August 13, 2026 that the federal Wiretap Act claim, the California claim and two Illinois voiceprint claims could proceed (per recordinglaw.com). That is a ruling on the pleadings, not a finding of liability.

For an owner, the takeaway is practical rather than alarming: before you connect an AI tool to your phone line, find the clause that says what the vendor may do with call audio and transcripts, and prefer terms that limit use to serving you. The Avoma checklist makes the same point in the notetaker context, listing no-training clauses and SOC 2 among its vendor questions.

Google Ads and Local Services Ads: call recording you may already have switched on

Google Ads call recording is, per Google's help page, enabled by default to allow AI-powered lead qualification. The exceptions are accounts that previously disabled it and healthcare and financial-services advertisers. Callers hear an automated message that the call is recorded 'for quality purposes.' Advertisers must accept the Call Ads Supplemental Terms and acknowledge that they have given notice to agents, employees and others who may be on calls. It applies to US and Canadian numbers. To turn it off, go to Admin > Account settings > Call ads > Call recording > Off. Recordings are available for 30 days.

Search Engine Journal (April 21, 2026) advises advertisers to review whether Google's message satisfies their own legal obligations. A third-party post (ALM Corp) says the default starts July 1, 2026; Google's page shows no date, so check your own account setting rather than relying on a date.

For Local Services Ads, Google's data page says communications through Google may be monitored or recorded. Retention in North America is up to 180 days, or 60 days for legal verticals. Providers may record independently under the provider's company's privacy policies, and Google does not monitor or control how providers use the recordings. Providers are responsible for deleting consumer data stored on their own systems. Older secondary coverage says two years' retention; Google's current page says 180 days.

None of the other published AI-tool checklists reviewed here covers this, even though a local service business is likely to run Google ads.

Google Business Profile, Maps reviews and AI content

Business Profile no longer stores call history for you. Chat and call history were removed from Business Profile on July 31, 2024, and past records could be downloaded through Google Takeout until August 30, 2024. Calls still reach the listed phone number, and text and WhatsApp remain available for eligible accounts. If you want a record of customer contacts, you hold it in your own systems and carry the privacy duty that goes with it.

Google's Maps user-generated content policy applies to all Maps content, regardless of whether it is human-generated or AI-generated. Fake engagement is prohibited, including paid reviews and content not based on a real experience. Several blogs state that Google explicitly prohibits AI-generated review text. The policy wording says AI-generated content is held to the same rules; it does not say AI-written reviews are banned outright, so check before repeating that claim to staff or customers.

For website copy, Google Search Central's guidance on generative AI content (last updated December 10, 2025) says using generative AI to generate many pages without adding value for users may violate Google's spam policy on scaled content abuse. Google emphasizes accuracy, quality and relevance, and transparency about how content was created. Google's spam policies (last updated 2026-08-28) define scaled content abuse as many pages generated for the primary purpose of manipulating search rankings and not helping users, and name generative AI as one way to do it. Google's guide to optimizing for generative AI features (last updated 2026-07-10) says structured data isn't required for generative AI search and no special schema.org markup is needed. It adds that Merchant Center feeds and Google Business Profiles can help your products and services be visible in both AI responses and other Google Search results, and it describes 'Business Agent' as a conversational experience on Google Search that helps customers chat with your brand.

None of these Search Central pages addresses customer-data privacy or call-recording consent; that is outside their scope.

Text-message consent: the TCPA rules in plain terms

Under the TCPA, texts count as calls. Marketing texts need prior express written consent. Keep a record of the date and time, the method of consent, and the exact wording the customer saw (ActiveProspect). A sample disclosure many vendors use reads: 'Consent is not a condition of purchase. Message frequency varies. Message and data rates may apply. Reply STOP to cancel,' with links to your terms and privacy policy (SimpleTexting). Treat that as an example of the pattern, not as approved legal text.

The Eleventh Circuit vacated the FCC's 'one-to-one consent' rule on January 24, 2025 in Insurance Marketing Coalition v. FCC.

Revocation of consent (47 CFR § 64.1200(a)(10)), per Hunton's January 12, 2026 summary, has two parts. In effect since April 11, 2025: honor a revocation within a reasonable time not to exceed 10 business days. The words STOP, QUIT, END, REVOKE, OPT OUT, CANCEL and UNSUBSCRIBE are per se reasonable, and any other reasonable method counts as well. Delayed: the 'revoke-all' provision, under which an opt-out to one message type applies to all robocalls and robotexts from that caller. FCC order DA 26-12 of January 6, 2026 moved it from April 11, 2026 to January 31, 2027 (Burr & Forman). The FCC order text itself could not be read; this date comes from law-firm summaries. Plan for revoke-all anyway: if a customer opts out of promotions, assume you may soon need to stop all automated texts to them.

Quiet hours: telephone solicitations to residences are barred before 8 a.m. and after 9 p.m. local time of the recipient, under 47 CFR § 64.1200(c)(1), and this applies to texts (Federalist Society commentary).

AI-generated voices on calls are 'artificial' under the TCPA per the FCC declaratory ruling of February 8, 2024, so they need prior express consent.

Business texting registration (A2P 10DLC) and carrier fees

Texting customers from a business number through a provider such as Twilio involves registering your brand and campaign for A2P 10DLC. Twilio's help center, as seen in search snippets (the page itself could not be fetched), states that brand registration is $4.50 one-time for low-volume standard and sole-proprietor brands, raised from $4.00 on August 1, 2025. The campaign vetting fee is $15, and the monthly campaign fee is about $1.50 for low-volume mixed use.

Twilio's US SMS pricing page, marked current as of August 2026, lists $0.0083 per SMS segment, plus carrier surcharges per message of $0.0035 (AT&T), $0.0045 (T-Mobile), $0.0045 (Verizon) and $0.005 (US Cellular).

Twilio's Authentication+ help page says brands not verified by January 30, 2026 can't register new campaigns. It applies to public brands, so check whether it applies to you.

None of the other AI-tool checklists reviewed covers 10DLC registration or its costs.

Twilio texting fees for a small US sender

ItemAmountNote

Brand registration (low-volume standard and sole-proprietor)

$4.50 one-time

Raised from $4.00 on August 1, 2025; from search snippets of Twilio's help page

Campaign vetting fee

$15

From search snippets

Monthly campaign fee (low-volume mixed use)

About $1.50

From search snippets

SMS segment

$0.0083

Twilio US SMS pricing, current as of August 2026

Carrier surcharge, AT&T

$0.0035 per message

Twilio US SMS pricing

Carrier surcharge, T-Mobile

$0.0045 per message

Twilio US SMS pricing

Carrier surcharge, Verizon

$0.0045 per message

Twilio US SMS pricing

Carrier surcharge, US Cellular

$0.005 per message

Twilio US SMS pricing

AI disclosure laws that reach small businesses

Maine (Public Law 2025, ch. 294; 10 M.R.S. § 1500-DD, effective September 24, 2025) requires clear notice that a customer is not talking to a human when a chatbot could mislead them. A violation is an Unfair Trade Practices Act violation. The bill text is LD 1727 on the Maine Legislature site; the effective date and citation come from secondary summaries.

Utah's AI Policy Act (SB 149, amended by SB 226 and SB 452) requires disclosure when a user asks, and in some cases earlier. That description comes from a secondary source only, and the Utah code page could not be loaded, so read the current statute before relying on it.

The same secondary page concludes that only Maine and Utah reach ordinary small-business chatbots. That has not been verified against each state's text. It also mentions California SB 243, which concerns companion chatbots; no claim that a California customer-service-bot law covers only $500M+ companies could be confirmed.

Even where no statute requires it, telling callers and chat users they are dealing with AI is consistent with the transparency Google emphasizes for AI content and with the Dialzara checklist's suggestion to disclose within 30 seconds.

Penalty and exposure figures

The figures below show why consent records and vendor terms matter. They are statutory amounts, not predictions of what any one business would pay.

Statutory exposure figures

LawWhat it coversExposureCaveat

TCPA private right of action (47 U.S.C. § 227(b)(3))

Calls and texts made without required consent

$500 per violation, up to $1,500 if willful or knowing

Texts count as calls; statute text not opened, figures per Wikipedia overview and law-firm pages

FTC Do Not Call and Telemarketing Sales Rule

Telemarketing violations

Civil penalty up to $53,088 per violation (2025 adjustment)

Per FTC guidance and Federal Register notice, January 17, 2025

California Invasion of Privacy Act (Penal Code § 632)

Recording confidential communications without consent

$5,000 per violation in civil damages

Per recordinglaw.com, which lists other state penalties in the same table

CCPA (California privacy law)

Businesses meeting the thresholds

Administrative fines up to $2,663 per violation, or $7,988 if intentional or involving minors under 16; statutory damages $107 to $799 per consumer per incident

Revenue threshold $26,625,000 as of Jan 1, 2025 (CPPA notice); the California AG page still shows the older $25 million. Also applies at 100,000+ California consumers or households, or 50%+ of revenue from selling or sharing personal information

Illinois BIPA

Biometrics, including voiceprints

$1,000 per negligent and $5,000 per intentional or reckless violation

A 2024 amendment (SB 2979) limits recovery to one per person

Security basics from the FTC, applied to AI tools

The FTC's Cybersecurity for Small Business page (dated September 2025) does not mention AI, but its advice maps directly onto AI vendors. It says to include provisions for security in your vendor contracts, to limit vendor access to the data points they need, and to establish processes to confirm that vendors follow your rules, adding 'Don't just take their word for it.' It advises requiring multi-factor authentication for network access and for areas holding sensitive data. After a compromise, notify affected people and point them to IdentityTheft.gov.

The FTC's Protecting Personal Information guide for business frames data protection as five steps: take stock, scale down, lock it, pitch it, plan ahead. For AI tools that means knowing which tools hold customer data, keeping only what you need, restricting access, deleting data you no longer need (set retention windows), and having an incident plan.

Consumer versus business AI accounts: several secondary guides (Intellspot, Centre Technologies) say consumer ChatGPT plans (Free, Plus, Pro) may use conversations for training by default, while business plans do not. OpenAI's own enterprise privacy page and data controls FAQ could not be opened to confirm the wording, so verify each vendor's terms yourself and keep customer information out of personal accounts.

What generic AI-checklists leave out for a local service business

Published AI voice and notetaker checklists are useful for vendor-evaluation questions but leave gaps that matter to an owner-operator. None addresses Google Local Services Ads or Google Ads default call recording. None mentions the TCPA revocation timeline, including the January 31, 2027 delay. None covers 10DLC registration or its costs. None covers the 2026 Otter ruling or the Ambriz vendor-capability theory, both of which turn on the vendor's data-use clause. None gives plan prices for the receptionist tools they discuss. Most are written for contact centers or enterprise IT, and several repeat unsourced counts of all-party states that disagree (11, 12, 13).

Use them as a supplement, not a substitute, and treat their statistics with care: some are unsourced, and one cites a 2023 breach average of $9.48M.

Other checklists and what each covers

PageDateCoversMisses or weaknesses

Dialzara, AI Phone Agent Compliance Checklist

May 25, 2026

Disclosure within 30 seconds, outbound consent, 10-business-day revocation, retention, TLS 1.2+/AES-256/MFA, vendor-contract items, penalty figures (TCPA $500 to $1,500, DNC $53,088, CCPA, HIPAA, PCI)

No TCPA text/SMS specifics, no Google guidance, no contract wording examples; cites a 2023 breach average ($9.48M)

AInora, AI Voice Agent Security Checklist

Apr 3, 2026

20 items across encryption, access, consent, retention, incident response, vendor assessment

Statistics largely unsourced and GDPR-heavy; no 10DLC, TCPA or text consent; does not name the two-party states

Avoma, AI Notetaker Security Checklist

Undated (references Aug 2025)

20 vendor-evaluation questions, SOC 2, no-training clauses, the Otter lawsuit

Aimed at meeting notetakers and enterprise IT (SAML SSO); no SMS, no state recording-law detail, nothing for owner-operators

CommLaw Group, 7 Legal Tips

Mar 14, 2025

TCPA, AI disclosure laws, CIPA, BIPA-type issues, call-recording states

No statistics, costs or checklist; predates the 2026 rulings

Vida, Call Recording Compliance Guide

Jun 29, 2026

Setup and compliance walkthrough

Unsourced statistics; no pricing; vendor promotion

Frontdesk, Call Recording Compliance

Jul 9, 2025

Consent, storage, retention framework

Unsourced stats; no state list detail; no fine amounts

Research context

Facts to keep in view

Ransomware in small-business breaches: 88%

Verizon's 2025 Data Breach Investigations Report (published April 23, 2025) found ransomware in 44% of all breaches but in 88% of breaches involving small and medium businesses. The 88% figure comes from Verizon's press release summary.

Third-party involvement in breaches: 30%, doubled

Verizon's 2025 report says third-party involvement in breaches doubled to 30%. Every AI vendor you connect to customer data is a third party.

FBI IC3 reported losses, 2025: $20.877B

The FBI's Internet Crime Complaint Center received 1,008,597 complaints, up 26% in losses on 2024. It also began tracking 'AI-related' complaints: 22,000+ complaints and nearly $900M in losses. Figures come from secondary summaries of the report released April 7, 2026.

TCPA private damages: $500 to $1,500 per violation

$500 per violation, up to $1,500 if willful or knowing. Texts count as calls. About 80% of TCPA suits are class actions.

Text opt-out deadline: 10 business days

Since April 11, 2025, a business must honor a revocation of text or call consent within a reasonable time not to exceed 10 business days. The broader 'revoke-all' provision was delayed to January 31, 2027.

Google Ads call recording: On by default

Google Ads call recording is enabled by default to allow AI-powered lead qualification, except for accounts that previously disabled it and healthcare and financial-services advertisers. Recordings are available for 30 days.

Local Services Ads recording retention: Up to 180 days

Google says communications through Local Services Ads may be monitored or recorded and retained up to 180 days in North America (60 days for legal verticals). Google does not control how providers use their own recordings.

All-party recording states: 11 to 13, depending on the count

Federal law is one-party consent, but published lists of all-party states disagree, mainly over Connecticut, Michigan, Oregon, Vermont and Delaware. If either party is in an all-party state, get everyone's permission.

10DLC brand registration (Twilio): $4.50 one-time

For low-volume standard and sole-proprietor brands, raised from $4.00 on August 1, 2025. Campaign vetting is $15, and the monthly campaign fee is about $1.50 for low-volume mixed use. These amounts come from search snippets of Twilio's help page.

A practical path

What to do

  1. 01

    Take inventory. List every tool that hears, reads or stores customer information: AI receptionist, chatbot, call recorder, note-taker, texting platform, ad platform recording. The FTC's data framework starts with 'take stock', then 'scale down', 'lock it', 'pitch it' and 'plan ahead'.

  2. 02

    Check which account type you are on. Several guides say consumer ChatGPT plans (Free, Plus, Pro) may use conversations for training by default while business plans do not. Confirm each vendor's current terms yourself; OpenAI's own pages could not be verified.

  3. 03

    Read the vendor's data-use clause. The legal theories in Ambriz v. Google and In re Otter.AI turn on a vendor retaining and using conversations for its own purposes, such as model training. Ask whether the vendor can, and does, use your customers' calls for anything but serving you.

  4. 04

    Put security terms in the contract and limit access. The FTC advises security provisions in vendor contracts, giving vendors access only to the data points they need, and confirming that vendors follow your rules rather than taking their word for it.

  5. 05

    Turn on multi-factor authentication for your network and for any area holding sensitive data, including your AI, phone and texting dashboards.

  6. 06

    Decide your recording-consent approach. If either party to a call may be in an all-party-consent state, get everyone's permission. A recorded announcement at the start of every call is the simplest way to apply one rule everywhere.

  7. 07

    Check your Google settings. In Google Ads, confirm whether call recording is on and whether Google's automated message meets your own legal obligations (Admin > Account settings > Call ads > Call recording). If you run Local Services Ads, know that Google may record and retain communications up to 180 days and that you are responsible for deleting data on your own systems.

  8. 08

    Set retention and deletion rules. Know how long each tool keeps recordings and transcripts. Some plans sell retention as a feature (Goodcall's Starter plan keeps call details 7 days, Growth 30 days, Scale unlimited), so choose the shortest window that serves your business.

  9. 09

    Capture text consent properly. For marketing texts, get prior express written consent and keep the date and time, the method, and the exact wording shown to the customer.

  10. 10

    Honor opt-outs fast. Treat STOP, QUIT, END, REVOKE, OPT OUT, CANCEL and UNSUBSCRIBE as valid revocations, and stop within 10 business days at most. Send solicitation texts to residences only between 8 a.m. and 9 p.m. in the recipient's local time.

  11. 11

    Register your business texting. Complete A2P 10DLC brand and campaign registration with your texting provider before you send volume, and check whether the Twilio authentication requirement for public brands (deadline January 30, 2026) applies to you.

  12. 12

    Disclose AI where required. If you use an AI voice for outbound calls, you need prior express consent. If you use a customer-facing chatbot, tell customers it is not a human, which Maine requires when a chatbot could mislead them and Utah requires in certain cases.

  13. 13

    Write a breach plan. If customer data is compromised, notify affected people and point them to IdentityTheft.gov, as the FTC advises.

Watch for these

Common mistakes

Assuming federal one-party consent covers every call

Federal law is one-party consent, but California and other all-party states apply their own rules. California law applies to calls to or from California regardless of where the caller is. If either party is in an all-party state, the safest approach is to get everyone's permission.

Trusting a vendor's marketing instead of its data-use clause

The wiretap cases against Google's call-center AI and Otter.AI turn on whether a vendor has the capability to use conversations for its own purposes, such as model training. The FTC advises you to confirm that vendors follow your rules and not just take their word for it.

Leaving Google Ads call recording on without reviewing it

Google enables call recording by default for AI-powered lead qualification and plays callers a message that calls are recorded 'for quality purposes.' Search Engine Journal advises advertisers to check whether that message satisfies their own legal obligations.

Using a personal or consumer AI account for customer information

Several guides say consumer ChatGPT plans may use conversations for training by default and business plans do not. Verify each vendor's current terms before pasting customer details into any tool.

Treating a text opt-in as permanent or undocumented

You must stop texting within 10 business days of a revocation, and marketing texts need prior express written consent backed by a record of the date and time, method and exact wording shown. A broader revoke-all rule takes effect January 31, 2027.

Using an AI voice for outbound calls without consent

The FCC ruled on February 8, 2024 that AI-generated voices are 'artificial' under the TCPA, so these calls need prior express consent.

Assuming Google stores your call history

Chat and call history were removed from Business Profile on July 31, 2024. Calls still reach your listed number, but if you want a record, you need to keep it in your own systems, and then you are responsible for protecting and deleting it.

Repeating the claim that Google bans AI-written reviews outright

Google's Maps policy says its rules apply to user-generated content whether human- or AI-generated, and it prohibits fake engagement including paid reviews and content not based on a real experience. The policy wording does not say AI-written reviews are banned outright.

Choose with context

A decision guide

  1. Recording a phone call

    What the rule requires
    Federal law is one-party consent. All-party states require everyone's permission, and the safest approach is to get it whenever either party may be in one.
    Key date or caveat
    Published counts of all-party states range from 11 to 13. California law applies to calls to or from California.
  2. Placing a call with an AI-generated voice

    What the rule requires
    Prior express consent under the TCPA (FCC declaratory ruling).
    Key date or caveat
    Ruling dated February 8, 2024.
  3. Sending marketing texts

    What the rule requires
    Prior express written consent, with a record of date and time, method and exact wording shown.
    Key date or caveat
    Texts count as calls under the TCPA.
  4. Honoring STOP or other opt-outs

    What the rule requires
    Stop within a reasonable time not to exceed 10 business days. STOP, QUIT, END, REVOKE, OPT OUT, CANCEL and UNSUBSCRIBE are per se reasonable; other reasonable methods also count.
    Key date or caveat
    In effect since April 11, 2025. The revoke-all provision moved to January 31, 2027.
  5. Texting or calling residences for solicitation

    What the rule requires
    Not before 8 a.m. or after 9 p.m. in the recipient's local time.
    Key date or caveat
    Applies to texts.
  6. Using a customer-facing chatbot

    What the rule requires
    Maine: clear notice that the customer is not talking to a human when a chatbot could mislead them. Utah: disclosure when asked, and in some cases earlier.
    Key date or caveat
    Maine's law took effect September 24, 2025. Utah's details come from secondary sources, so check the text.
  7. Capturing voiceprints in Illinois

    What the rule requires
    Voiceprints are covered by the Illinois biometric law (BIPA).
    Key date or caveat
    A 2024 amendment limits recovery to one per person.
  8. Running Google Ads call tracking in the US or Canada

    What the rule requires
    Google records by default unless you turned it off; you accept the Call Ads Supplemental Terms and acknowledge you have given notice to agents and others on calls.
    Key date or caveat
    Healthcare and financial-services advertisers are excluded from the default.

Takeaway

For a local service business, the highest-value checks are few and concrete: read each AI vendor's data-use clause, record only with consent that satisfies the strictest state involved, review the Google Ads call-recording setting, and keep dated records of text consent while honoring opt-outs within 10 business days. Everything else in this guide supports those four habits.

  • Before connecting an AI receptionist or note-taker

    Read the vendor's data-use clause

    The Ambriz v. Google and In re Otter.AI claims turn on a vendor's capability to use conversations for its own purposes, and the FTC advises confirming that vendors follow your rules rather than taking their word for it.

  • Before you record any call

    Announce recording on every call

    Federal law is one-party consent, but if either party is in an all-party state the safest approach is to get everyone's permission, and published counts of those states range from 11 to 13.

  • If you run Google Ads or Local Services Ads

    Check your call-recording setting

    Google Ads recording is on by default for most accounts, and Local Services Ads communications may be retained up to 180 days, with you responsible for data on your own systems.

  • Before texting customers

    Document consent and honor STOP within 10 business days

    TCPA damages run $500 to $1,500 per violation, and a revoke-all rule takes effect January 31, 2027.

Questions

Frequently asked

Do I need consent to record customer calls with an AI receptionist?
Federal law requires consent from one party. All-party-consent states, listed variously as 11, 12 or 13 depending on the source, require everyone on the call to agree. If either party may be in an all-party state, the safest approach is to get everyone's permission, for example with an announcement at the start of every call. California law applies to calls to or from California regardless of where the caller is.
Which states require all parties to consent to a recording?
Lists differ. One source lists California, Delaware, Florida, Illinois, Maryland, Massachusetts, Montana, Nevada, New Hampshire, Pennsylvania and Washington. CommLaw Group adds Connecticut. Recordinglaw.com counts 13, adding Connecticut and Oregon with caveats. The disagreement concerns Connecticut, Michigan, Oregon, Vermont and Delaware, so confirm your own state and your customers' states against current law.
Does Google record my Google Ads calls?
Google Ads call recording is enabled by default to allow AI-powered lead qualification, except for accounts that previously disabled it and healthcare and financial-services advertisers. Callers hear that the call is recorded for quality purposes, and recordings are available for 30 days. To turn it off, go to Admin > Account settings > Call ads > Call recording > Off. Review whether Google's message satisfies your own legal obligations.
How long does Google keep Local Services Ads communications?
Google's Local Services Ads data page says retention in North America is up to 180 days, or 60 days for legal verticals. Older coverage says two years, but Google's current page says 180 days. Google also says it does not control how providers use their own recordings, and providers are responsible for deleting consumer data stored on their own systems.
What do I need before sending marketing texts to customers?
Prior express written consent, with a record of the date and time, the method, and the exact wording shown to the customer. Honor STOP, QUIT, END, REVOKE, OPT OUT, CANCEL and UNSUBSCRIBE within 10 business days at most, and do not send solicitations to residences before 8 a.m. or after 9 p.m. in the recipient's local time. Register for A2P 10DLC through your texting provider.
When does the TCPA revoke-all rule take effect?
January 31, 2027. FCC order DA 26-12 of January 6, 2026 moved it from April 11, 2026. Under it, an opt-out to one type of message applies to all robocalls and robotexts from that caller. The 10-business-day revocation rule has been in effect since April 11, 2025.
Can I use an AI voice to call customers?
AI-generated voices are 'artificial' under the TCPA according to the FCC's February 8, 2024 declaratory ruling, so these calls need prior express consent.
Do I have to tell customers they are talking to a chatbot?
Maine requires clear notice that a customer is not talking to a human when a chatbot could mislead them, effective September 24, 2025. Utah's AI Policy Act requires disclosure when a user asks and in some cases earlier. The Utah details come from a secondary source, so check the statute. Other states may have different rules.
Is it safe to paste customer information into ChatGPT?
Several secondary guides say consumer ChatGPT plans (Free, Plus, Pro) may use conversations for training by default while business plans do not. OpenAI's own pages could not be opened to confirm that wording, so read the current terms for the exact plan you use and keep customer information out of personal accounts.
What does a data breach cost a small local business?
No primary-source average for a local service business under 50 employees has been found. IBM's 2026 global average across organizations is $4.99M, and secondary coverage puts the US average at $11.5M, but these describe organizations in general. Verizon's 2025 report found ransomware in 88% of breaches involving small and medium businesses, with a median ransom paid of $115,000.
Does Google ban AI-written reviews?
Google's Maps policy says its rules apply to user-generated content whether human- or AI-generated, and it prohibits fake engagement, including paid reviews and content not based on a real experience. Several blogs claim it explicitly bans AI-generated review text, but the policy wording does not say that.
Is the Otter.AI case a finding that AI note-takers are illegal?
No. On August 13, 2026, Judge Eumi Lee let the federal Wiretap Act claim, the California claim and two Illinois voiceprint claims proceed. That is a ruling on the pleadings, not a finding of liability. The alleged problem is a vendor retaining and using conversations for its own purposes.

Keep reading

Related

63 sources, captured 2026-10-01 · show
  1. Verizon 2025 Data Breach Investigations Report press release
  2. Security Magazine on Verizon 2025 DBIR
  3. IBM Cost of a Data Breach
  4. Security Boulevard on IBM 2026
  5. eSecurity Planet on IBM 2026
  6. Kiteworks on IBM 2025
  7. Jones Walker on IBM 2025
  8. FBI IC3 2025 report
  9. SpyCloud on FBI IC3 2025
  10. Hoodline on FBI IC3 2025
  11. TCPAWorld on TCPA class action spike
  12. NatLawReview on TCPA filings
  13. ActiveProspect on TCPA lawsuits
  14. ActiveProspect on SMS consent
  15. Smith.ai pricing
  16. Goodcall pricing
  17. Ruby pricing
  18. Twilio US SMS pricing
  19. Twilio, Pricing and Fees for A2P 10DLC
  20. Twilio Authentication+ for public brand A2P 10DLC
  21. Wikipedia, Telephone Consumer Protection Act of 1991
  22. FTC, Complying with the Telemarketing Sales Rule
  23. Federal Register, civil penalty adjustments (Jan 17, 2025)
  24. CPPA CCPA threshold adjustment
  25. California AG, CCPA
  26. Greenberg Traurig on BIPA update
  27. Alston on BIPA damages limit
  28. Google, Guidance on Generative AI Content
  29. Google, Spam policies
  30. Google, Optimizing your website for generative AI features
  31. Google Search Central blog, May 2026
  32. Business Profile Help, chat and call history changes
  33. Google Maps user-generated content policy
  34. Google Maps fake engagement policy
  35. Google Ads, Set up call recording
  36. Google Ads, Play your call recordings
  37. Search Engine Journal on Google Ads call recording default
  38. ALM Corp on Google Ads call recording default
  39. Google Local Services Ads data page
  40. Viirtue, call recording consent laws by state
  41. CommLaw Group, 7 legal tips on AI in customer service
  42. Frontdesk, call recording compliance
  43. AInora, AI voice agent security checklist
  44. Dialzara, AI phone agent compliance checklist
  45. Recordinglaw.com, all-party consent states
  46. Recordinglaw.com, Otter.AI wiretap lawsuit
  47. Goodwin on AI voice products and CIPA
  48. ZwillGen on Ambriz v. Google
  49. Avoma, AI notetaker security checklist
  50. Vida, call recording compliance guide
  51. Eleventh Circuit, Insurance Marketing Coalition v. FCC
  52. Hunton on TCPA revocation rules
  53. Burr & Forman on TCPA revoke-all delay
  54. Federalist Society on TCPA quiet hours
  55. SimpleTexting, opt-in text disclaimer language
  56. FCC AI voice declaratory ruling
  57. Cooley on FCC AI robocall ruling
  58. Maine Legislature, LD 1727
  59. SiteGPT, AI chatbot compliance guide
  60. FTC, Cybersecurity for Small Business
  61. FTC, Protecting Personal Information
  62. Intellspot, AI privacy checklist
  63. Centre Technologies on ChatGPT guidelines

Vendor pages change. Figures are as published on the capture date; confirm before deciding.

Get your free Growth Brief