The Legal Definition: Consent, Timing, and Content

A single unwanted text can cost a business $500. The next one can cost $1,500. And the hundred thousandth can trigger class-action liability that industry analysis places above $150 million.

That is the regulatory reality behind the word "inappropriate." Under the Telephone Consumer Protection Act, inappropriate texting is not a matter of taste or tone. It is a specific legal breach, and it falls into three categories.

The TCPA requires prior express written consent before marketing texts go out. Without that documented permission, the message itself becomes the violation, regardless of how polite or relevant the content is.

Recent litigation has complicated this standard. The Fifth Circuit's ruling in Bradford v. Sovereign Pest Control now accepts oral consent in that circuit, but Holland & Knight attorneys Paul Bond and Kristen N. Ricci caution that oral consent "must still demonstrate that the called party provided clear, direct and unequivocal consent" and should be "carefully documented and independently verifiable to withstand future scrutiny." Written consent remains the safest standard across jurisdictions.

Federal rules prohibit texts before 8 a.m. or after 9 p.m. in the recipient's local time zone, according to ActiveProspect's compliance guide. States layer additional restrictions on top of that baseline:

• Florida limits messages to three per 24 hours and offers a 15-day safe harbor after opt-out.
• Virginia requires opt-out records to be retained for 10 years.
• Connecticut penalties can reach $20,000 per violation.

For multi-location organizations, these variations mean a campaign configured for one state can quietly violate the law in another.

Since April 2025, businesses must honor opt-outs through "any reasonable method." A reply saying "leave me alone," "remove me," or even a frustrated misspelling counts as a valid revocation, the keyword "STOP" is only one of many acceptable forms. The processing deadline is 10 business days, with a best-practice confirmation window of 5 minutes.

Failing to honor these requests converts a compliant campaign into a liability. TCPA class actions filed through mid-2025 increased by nearly 95% year-over-year, according to messaging compliance research, and settlements have ranged from $4.42 million (DSW, March 2025) to $20 million (Uber, 2017).

The distinction matters. A text can be perfectly worded and still be inappropriate if it lands without consent, outside permitted hours, or after a revocation request. Regulators and courts do not weigh whether the message was helpful or annoying; they ask whether the sender followed the rules.

That is why list discipline sits at the center of compliant outbound operations. Checking list source and consent records before any campaign launches, flagging bought lists without clear permission records, and declining campaigns that cannot meet the standard, these are the safeguards that keep a campaign from crossing the line into statutory damages. The legal definition is precise, and the financial exposure for missing it is real.

Why AI Voices Trigger Stricter Compliance Standards

If your AI system sends a message without checking consent, regulators treat it the same as a robocall, and the penalties stack up just as fast. The FCC's February 2024 declaratory ruling made that clear: AI-generated voices are classified as "artificial voices" under the TCPA, which means they fall inside the same consent and registration framework as traditional automated calls and texts (Infobip).

That classification matters because it removes any argument that AI-assisted messaging is somehow a gray area. Prior express consent is required. Registration rules apply. And the same statutory damages of $500 per violation, or $1,500 for willful violations, hang over every message with no cap on total liability. A campaign of 100,000 non-compliant messages could expose a business to more than $150 million in class-action liability.

The bigger AI-specific risk, though, is opt-out handling. Since April 2025, FCC rules require businesses to honor opt-outs made through "any reasonable means," which goes well beyond the keyword STOP. That includes phrases like "stop texting me," "remove me," or even misspellings and frustrated language (Voxie). Older AI systems built on simple keyword matching miss these, and every missed opt-out is a violation waiting for a plaintiff's attorney.

This failure to recognize natural language revocations has become a primary driver of recent class-action lawsuits. TCPA class actions filed through mid-2025 increased by nearly 95% year over year, and compliance-focused vendors now describe "fuzzy opt-out recognition" as a critical safeguard: AI that detects intent behind the words rather than matching exact keywords (MoEngage).

So what should you actually check for before letting an AI system message anyone?

• Does the AI recognize opt-out intent in plain language, beyond the word "STOP"?
• Does it log and process those requests within the 10-business-day deadline?
• Is there AI disclosure on every call or message, so recipients know they're talking to an automated system?
• Has the contact list been reviewed for consent records and reassigned numbers before launch?

That last point deserves emphasis. Consent from a phone number's previous owner doesn't transfer to the new owner, which makes reassigned numbers one of the most common sources of inappropriate texting. It's why list discipline comes before technology. At My AI Call Center, list source and consent records are checked before any campaign launches, and bought lists without clear permission records are flagged and, in most cases, declined.

The takeaway is simple: the FCC doesn't care whether a human or a model pressed send. If the voice is artificial, the rules are strict, and the AI needs to be built to respect them.

The Financial Exposure of Non-Compliant Campaigns

You think a single missed opt-out is an administrative glitch. It is actually a liability trap.

TCPA violations carry statutory damages of $500 per message for standard infractions. Willful or knowing violations double that to $1,500 per message. There is no cap on aggregate liability. A campaign of 100,000 non-compliant texts could result in over $150 million in liability, according to industry compliance data.

This exposure stems from two common operational failures. The first is content classification. Many businesses send appointment reminders or payment notices. These are transactional messages. Adding a discount offer or upsell to that message reclassifies it as marketing. Marketing requires Prior Express Written Consent (PEWC). If you do not have PEWC for that specific contact, you are in violation.

The second failure is list hygiene. Phone numbers get reassigned. Consent from a previous owner does not transfer to the new owner. If you send a marketing text to a reassigned number, you are texting someone who never consented. This is a direct TCPA violation.

Real-world settlements illustrate the scale of this risk:

• Wells Fargo paid $17.85 million in 2019 for unsolicited texts and calls.
• Uber settled for $20 million in 2017 over unwanted driver-recruitment SMS.
• DSW faced a $4.42 million lawsuit in March 2025 for unwanted marketing SMS.

These are not isolated incidents. TCPA class actions filed through mid-2025 increased by nearly 95% year-over-year. The legal landscape is tightening. Regulators are closing loopholes that allowed lead generators to use a single consumer consent to authorize multiple third-party sellers. Each seller must now obtain separate, specific consent.

You cannot manage this risk with manual spot-checks. You need systematic safeguards. My AI Call Center reviews list source and consent records before any campaign launches. We flag bought lists without clear permission records. In most cases, we decline them. This discipline prevents the massive liability spikes seen in the settlements above.

The distinction between transactional and marketing content is critical. A payment reminder is safe. A payment reminder with a "10% off next time" tag is not. Our script approval process enforces this separation. We ensure that one clear goal per campaign is maintained without accidental reclassification.

Reassigned numbers are equally dangerous. We scrub contact lists against the FCC’s Reassigned Numbers Database to create a safe harbor. This step is non-negotiable for high-volume campaigns. It ensures that your consent records match the current owner of the phone number.

State-specific rules add another layer of complexity. Florida limits messages to three per 24 hours. Virginia requires opt-out records to be retained for 10 years. Connecticut penalties can reach $20,000 per violation. Your system must respect these local variations automatically.

If you are running outbound campaigns, the cost of error is high. You need a partner who treats list discipline as a core feature, not an afterthought. We report what actually happened. We never invent metrics. We honor opt-outs immediately.

Plan your next campaign with a team that checks consent records before you spend anything. From 9¢ per connected minute, we help you run more useful calls without building a bigger call center.

Building a Compliance-First AI Communication System

The cheapest compliance safeguard is the one built in before launch, not bolted on after a lawsuit. TCPA statutory damages run $500 per message, or $1,500 for willful violations, with no cap on aggregate liability. A single campaign of 100,000 non-compliant texts could theoretically produce over $150 million in exposure, according to compliance analysis. That math is why the safeguards below are non-negotiable, not optional polish.

Fuzzy opt-out detection comes first. Since April 2025, FCC rules require honoring opt-outs made by "any reasonable means," which includes phrases like "stop texting me," "remove me," and even misspellings or frustrated language, the keyword STOP is merely one example. A keyword-matching system misses those revocations, and every missed one is a violation. AI systems need natural language processing that detects intent, logs the request immediately, and stops messaging well inside the 10-business-day processing deadline. Best practice is to send a confirmation within 5 minutes.

Next is Reassigned Numbers Database scrubbing. Consent belongs to the person, not the phone number. When someone abandons a number and a new owner picks it up, the previous owner's consent does not transfer, and texting that number is inappropriate by definition. Scrubbing every list against the FCC's Reassigned Numbers Database before launch creates a documented safe harbor. This is also where bought lists fail most often, which is why My AI Call Center flags lists without clear permission records during the list and consent review, and in most cases declines them before anything is spent.

Quiet hours sound simple until you operate across state lines. Federal rules prohibit texts before 8 am and after 9 pm in the recipient's local time zone, but several states enforce stricter "mini-TCPA" laws on top of that. Florida caps messages at three per 24 hours and provides a 15-day safe harbor after opt-out. Virginia requires opt-out records to be retained for 10 years. State penalties stack on federal exposure too: Connecticut fines can reach $20,000 per violation. Any system you run must adjust sending windows automatically by the recipient's location, not the sender's.

A few checks belong in every pre-launch review:

• Verify consent records exist, with date, method, and exact language captured, since even where oral consent may suffice, attorneys advise it be "clear, direct and unequivocal" and independently verifiable.
• Scrub against the Reassigned Numbers Database and document the scrub date.
• Keep transactional and marketing content strictly separated, because adding an upsell to an appointment reminder reclassifies the whole message as marketing and raises the consent bar.
• Configure state-specific quiet hours, day restrictions, and frequency caps by recipient location.
• Confirm AI disclosure language, since AI-generated voices are treated as "artificial voices" under the TCPA and require prior express consent.

The stakes keep rising. TCPA class actions filed through mid-2025 increased by nearly 95% year over year, per the same analysis, and settlements like Wells Fargo's $17.85 million show the downside is real. This is the logic behind My AI Call Center's managed model: nothing launches until the list, consent records, script, and opt-out handling are approved, and every opt-out is logged and honored immediately. If your current process can't answer those five questions with documentation, the list isn't ready. If you'd rather have that review done for you, campaigns start with a free review at 9¢ per connected minute.