AI Voice for Charity Fundraising: PECR-Compliant at Scale

The scale problem every charity fundraising team knows

UK charities depend on telephone fundraising. Pledge upgrades, lapsed donor reactivation, Gift Aid conversion, thank-you calls — these are the call types that keep monthly giving programmes alive, sustain donor relationships, and drive the recurring income that funds the mission.

The problem is scale. A mid-size charity with 200,000 active and lapsed donors cannot call that list with a team of twenty telephone fundraisers. At best, they reach 3,000-4,000 contacts per week. Lapsed donors drift irretrievably away. Pledge upgrade windows close. Gift Aid declarations that could unlock tens of thousands in tax reclaim go uncaptured. The team is working hard; the programme is losing ground.

AI voice changes that arithmetic. The same programme that reaches 4,000 contacts per week with twenty agents can reach 40,000 or more with AI voice, at a fraction of the per-contact cost. Donor reactivation programmes running AI voice in 2025-2026 are recovering Gift Aid-eligible regular givers at a cost-per-reactivation of £4-10, against a human agency cost of £35-80 for the same outcome.

But charities carry more regulatory exposure than most outbound callers. PECR Regulation 21, the Telephone Preference Service, the Fundraising Preference Service, the Fundraising Regulator's Code of Fundraising Practice — and the ICO's heightened enforcement posture on automated calls — mean that a charity deploying AI voice without a compliance architecture is not just risking a fine. It is risking the reputational damage that ICO enforcement brings for a mission-driven organisation.

This guide maps the regulatory landscape for UK charity outbound calling, shows how AI voice improves rather than worsens compliance when built correctly, and gives the practical architecture every fundraising team should require before go-live.

This guide is published by the team behind Dilr Voice — enterprise AI voice deployed across regulated sectors including healthcare, financial services, and the third sector. For strategic programme design, see DATS, our five-stage consulting methodology.

---

The regulatory landscape for charity telephone fundraising

PECR Regulation 21: the governing rule

The Privacy and Electronic Communications Regulations 2003 (PECR) Regulation 21 is the primary rule for live marketing calls to UK individuals. It prohibits a call for direct marketing purposes to a subscriber who is registered with the Telephone Preference Service (TPS) or has previously notified the caller that they object to receiving such calls.

PECR Reg 21 applies regardless of whether the call is made by a human or an AI voice agent. It applies whether the caller is a commercial business, a charity, or a political organisation. The regulation does not recognise charitable status as a basis for exemption.

The critical point for charities: the "soft opt-in" under PECR Regulation 22 applies only to electronic mail — email, SMS, and equivalent automated messages. It does not apply to live telephone calls under Regulation 21. For outbound calls, the rule is binary: either the subscriber has not registered with TPS and has not previously objected, or you may not call.

What Regulation 21 does allow is calling non-TPS-registered numbers where there is no prior objection. This means:

• Existing donors who have not opted out and are not TPS-registered — legally callable under PECR, provided there is also a lawful GDPR basis (typically Legitimate Interests for existing supporters, subject to the requisite balancing test and LIA)
• Prospective donors (cold prospecting) — legally callable only if they are not TPS-registered and have not otherwise objected
• Anyone who has previously said "do not contact me" — may not be called, regardless of any other consideration

The Telephone Preference Service (TPS) and CTPS

The TPS is the UK's opt-out register for live marketing calls to personal numbers. The CTPS covers registered business numbers. PECR requires callers to check both lists before making any outbound marketing call and to suppress any registered number.

As of 2026, more than nine million UK numbers are registered with the TPS — a number that has grown significantly following ICO enforcement actions against both commercial and charity callers. Any outbound programme not running pre-call TPS scrubs is statistically certain to breach PECR on a regular basis.

Key scrubbing requirements:

• Must scrub against TPS/CTPS within 28 days of the planned call date
• Must maintain evidence of each scrub (date, file count, number of suppressions)
• Must apply suppression to every number in the campaign list, not just new additions since the previous campaign

The Fundraising Preference Service (FPS)

The FPS is a sector-specific register run by the Fundraising Regulator. It allows any member of the public to stop receiving contact from charities — across telephone, email, and post — either from all charities or from specific named organisations.

The FPS is not legally mandated under the same statutory framework as TPS, but signatories to the Code of Fundraising Practice — which includes virtually all UK charities above a certain size — are required to respect FPS registrations. In practice, disregarding FPS is a Fundraising Regulator enforcement matter, not an ICO matter, but the reputational consequences for a charity facing a Fundraising Regulator referral are severe.

For telephone calls, the FPS check should be run before every campaign, in addition to the TPS scrub. The FPS database is updated in near real-time, making a close-to-call-date check essential.

The Fundraising Regulator's Code of Fundraising Practice

Section 11 of the Code covers telephone fundraising directly. Key requirements for AI voice deployments:

• Identification: the caller must clearly state who they are and which charity they represent at the beginning of the call
• Scripted opt-out mechanism: the individual must be told how to opt out of future calls, and any opt-out expressed must be honoured immediately
• No psychological pressure: the Code prohibits techniques that exploit vulnerability or apply undue pressure
• Caller ID transparency: the calling line identification (CLI) presented must be a real, contactable number — not spoofed or withheld
• AI disclosure: since the 2024 Code update, charities using automated systems must disclose this clearly to the call recipient at the first interaction

The AI disclosure requirement in the Code predates the EU AI Act Article 50 transparency obligation. For UK charities calling UK donors, the Fundraising Regulator Code is the operative disclosure rule. For charities calling EU-resident donors, Article 50 applies from 2 August 2026.

For the full UK outbound calling compliance architecture — covering PECR, GDPR, and ICO expectations beyond the charity sector — see our guide to AI outbound calling: GDPR and PECR compliance.

ICO enforcement context

The ICO can issue monetary penalties under PECR of up to £500,000, and increasingly routes enforcement through the UK GDPR power, which extends to £17.5 million or 4% of global annual turnover. The charity sector has seen significant enforcement activity:

• Multiple enforcement notices issued to charity telephone fundraising agencies between 2020 and 2025 for calling TPS-registered numbers
• Operation Trident (ICO, 2024) specifically targeted charity cold-calling compliance
• The ICO has signalled that automated calling systems attract additional scrutiny — an AI voice agent calling a TPS-registered number combines two high-priority enforcement areas in one event

For a mission-driven organisation, an ICO fine generates press coverage, donor trust questions, and in some cases major donor withdrawal that far exceeds the financial penalty itself. Compliance is not optional for charities — it is existential.

---

Why AI voice improves compliance posture — when built correctly

The instinct many fundraising teams have is that AI voice creates more regulatory risk. The reality, when the platform is configured correctly, is the opposite. Human telephone operations introduce compliance risk through inconsistency, fatigue, and process gaps that AI voice eliminates by design.

Systematic PECR compliance instead of human fallibility

Human telephone agents make mistakes. TPS registrations are overlooked. Suppression lists are applied incorrectly. Opt-outs recorded by one agent are not propagated to the shared list before another agent calls the same number. At scale, with multiple agents working from spreadsheet-based calling lists, PECR breaches happen not because the team is careless but because the manual system cannot hold the complexity required.

A properly configured AI voice platform executes pre-scrub checks automatically before a campaign dial starts. TPS and FPS suppression is applied to every number in the list, with an audit trail timestamped to the run. The suppression log is evidence — the kind ICO investigators expect to see when they arrive.

Consistent disclosure on every call

The Fundraising Regulator Code requires disclosure of AI or automated systems on every call. A human team can forget, rush, or phrase it inconsistently. A scripted AI voice system cannot: the opening disclosure is part of the system prompt and is executed on every call without variation.

The disclosure script format that satisfies both the Code and ICO guidance on automated calling:

"Hello, this is an automated call on behalf of [Charity Name]. I'm contacting [donor name] about [your regular gift / your past support / a short message from us]. You can speak to me naturally or say 'human' at any time to speak with an advisor instead."

This opening achieves in six seconds: identity disclosure, purpose disclosure, AI/automated-system disclosure, and an immediate opt-to-human escalation route — every element the Code requires.

Opt-out capture with immediate propagation

Under PECR and the Code, an opt-out expressed during a call must be respected immediately and for all future campaigns. In human calling operations, opt-outs are frequently recorded in call notes but not propagated to the CRM until end-of-day or end-of-week batch processing. In the interval, another agent may call the same number.

AI voice agents integrated with the charity's CRM write opt-out flags in real time — before the call has ended. The donor record is updated, the suppression list is augmented immediately, and the number cannot be scheduled for any subsequent outbound campaign.

Gift Aid consent capture at scale

Gift Aid is one of the highest-value activities a charity telephone programme can run. A valid Gift Aid declaration means the charity can reclaim basic-rate tax on donations — typically 25p for every £1 donated. Human callers capture Gift Aid consent inconsistently, and the evidence trail is often weak.

An AI voice agent integrated with the charity's Gift Aid processing system captures the declaration confirmation on every eligible call, writes the timestamp and verbal consent record to the donor file, and flags the record for the written confirmation letter that best practice requires. It does not forget to ask. The audit trail is complete from the first call.

For the full legal architecture of consent capture in AI voice calls — lawful basis, GDPR Article 6, and PECR interplay — see consent capture in AI voice calls: GDPR and PECR guide.

---

The four call types charities should automate

Not all charity call types are equal candidates for AI voice. The best fit is structured, high-volume interactions where the outcome is defined (yes/no, accept/decline, confirm/decline) and where the value is in reaching as many contacts as possible rather than in a single complex conversation.

1. Pledge upgrade calls

Existing regular donors giving £10 per month are often willing to increase to £12 or £15 if asked at the right moment with the right message. The upgrade ask is a short, structured call:

• Opening: name, charity, and purpose with AI disclosure
• Brief mission context (one sentence — what the charity has achieved recently)
• The ask: "Would you consider increasing your gift from £X to £Y per month?"
• If yes: payment confirmation and thanks
• If no: warm close, opt-out check, and thanks for their continued support

This is an archetypal AI voice call. The conversation flow is predictable, the outcome is binary, and the economics are compelling. Upgrading 500 donors at an average of £3 per month each adds £18,000 in annual giving. At AI voice costs of £0.60-£1.20 per contact, reaching 10,000 donors to achieve those 500 upgrades costs under £12,000. The same campaign with a human calling agency at £15 per contact costs £150,000.

2. Lapsed donor reactivation

Donors who gave regularly but stopped in the past 12-24 months represent the highest-ROI segment in the charity telephone programme. They carry an existing relationship — meaning they are legally callable under PECR if not TPS-registered — and they have demonstrated giving intent. The conversation is:

• Opening: name, charity, and that the charity values their previous support
• Brief update on impact since their last gift
• The ask: "Would you like to restart your regular gift?"
• If yes: gift amount, payment method, and Gift Aid capture
• If no: respectful close, opt-out check, and thanks

A lapsed file of 50,000 contacts that a human calling team takes six months to work through can be contacted by AI voice in a week. The economics at a 5-8% reactivation rate — typical for warm lapsed callers — deliver meaningful income at a cost-per-reactivated-donor that is a fraction of human agency rates.

3. Gift Aid conversion and annual re-confirmation

Many charities hold active donors for whom they do not have a valid Gift Aid declaration — either it was never captured, or the donor's tax situation has changed and needs re-confirmation. A Gift Aid conversion call is a pure administrative task:

• Confirm the donor pays income tax at least equal to the Gift Aid claimed
• Capture the verbal declaration with timestamp
• Confirm that a written follow-up will be sent by post or email

An AI voice agent handles this at scale. A charity running a Gift Aid audit across 80,000 donors without declarations can call the full list in under two weeks, capture declarations in real time, and create a timestamped audio record against every donor file. At 25p reclaim per £1 donated, a £10/month donor is worth an additional £30 per year in Gift Aid. Converting 20,000 donors to Gift Aid adds £600,000 in annual income.

4. Thank-you and relationship calls

Thank-you calls made within 48-72 hours of a donation increase donor retention. Research from the Fundraising Regulator's sector studies consistently shows that a personal contact following the first or a milestone gift reduces lapse rates by 15-30%. AI voice, configured to make a brief and warm thank-you call, provides this at a scale and cost that human operations cannot sustain.

The thank-you call is not an ask. It is:

• A genuine expression of thanks, by name, for the specific gift
• A brief statement of impact ("your support helps us...")
• An invitation to contact the charity if they have questions
• A reminder of how to amend or cancel their direct debit if needed

This last point — proactively offering the management option — is increasingly expected under the consumer-protection framing the Fundraising Regulator has been tightening. An AI voice programme that includes it in every thank-you call demonstrates proactive compliance.

---

Building the PECR-compliant architecture

A compliant AI voice fundraising programme has four layers. Each one has compliance obligations; each one also has a direct operational benefit.

Layer 1 — Suppression stack

Before a single call is made, every number in the campaign list must pass through:

1. TPS scrub (28-day currency): run against TPS API or a registered TPS data service
2. CTPS check (where any business numbers are included): same workflow
3. FPS check (near real-time): run against the Fundraising Regulator's FPS database before each campaign
4. Internal DNC list: every number from which an objection or opt-out has ever been received, maintained in the charity's CRM
5. Recent-contact check: many charities apply a 90-day cooling-off between calls to the same number; the AI voice platform applies this as a campaign filter

The AI voice platform should execute all five layers automatically from campaign configuration, producing a suppressed calling list and a timestamped audit record of the suppression run.

For the detailed DNC architecture that sits underneath this suppression stack — including real-time DNC checking within the call flow itself — see our guide to DNC logic in AI voice diallers.

Layer 2 — Disclosure and consent capture

Every call must begin with a compliant opening. The opening is part of the AI voice system prompt and executes on every call without exception. A Fundraising Regulator-compliant and ICO-compliant disclosure opening:

"This is an automated call from [Charity Name]. I'm calling [donor name] to talk about [your regular gift / your past support]. You can speak to me naturally at any time. If you'd prefer to speak with a human advisor, just say 'human' or press 1."

During the call, any consent given — Gift Aid declaration, pledge upgrade agreement, single-gift commitment — must be:

• Timestamped to the second against the call record
• Associated with the donor's phone number and CRM record
• Written to the CRM in real time before the call ends
• Flagged for a written follow-up confirmation (required for Gift Aid; best practice for pledge upgrades)

Layer 3 — Opt-out processing

Any expression of "please don't call again," "remove me from your list," "I don't want calls from charities," or equivalent must trigger:

1. Immediate call close — no further asks on the current call
2. A DNC suppression flag written to the donor record in real time
3. The number added to the internal DNC list with a timestamp
4. A suppression confirmation communication within 28 days (best practice; required under PECR for any written request to stop calling)

The AI voice agent's system prompt must recognise opt-out intent in natural language — not only scripted trigger phrases, but variations such as "take me off your list," "I'm not interested in charity calls," "please stop phoning me," or "I've already asked you not to call."

Layer 4 — Audit and evidence

ICO investigations arrive with requests for evidence. AI voice platforms generate that evidence automatically:

• Suppression logs: timestamped TPS, CTPS, FPS, and DNC suppression runs for each campaign
• Consent records: timestamped audio captures and CRM writes for every consent event
• Call recordings: stored against each donor record with access controls
• Opt-out processing records: every DNC flag, with timestamp and propagation confirmation
• Disclosure compliance records: system-level evidence that the disclosure opening was served on every call

A charity using AI voice for fundraising can respond to an ICO investigation request in hours. The equivalent evidence reconstruction from human calling operations typically takes weeks.

---

The ROI model for charity AI voice fundraising

Charities operate under close cost scrutiny. The proportion of income spent on fundraising is a public metric and a governance concern. AI voice significantly improves fundraising return on investment.

Pledge upgrade programme — indicative economics

The net income improvement from moving to AI voice for pledge upgrade calling, across a 100,000-contact donor base running two campaigns per year, is typically £80,000-200,000 in reduced campaign cost — before the income benefit of reaching more donors at the same budget.

Lapsed donor reactivation

• Human agency cost per reactivated donor: £35-80
• AI voice cost per reactivated donor: £4-12
• Retention differential: donors who receive a personal thank-you and relationship call within the first year are 15-30% more likely to remain active in the following 12 months (source: Fundraising Regulator sector research)

Gift Aid uplift

Gift Aid on a £10 per month donor is worth £30 per year to the charity — 25% reclaim on £120 of annual giving. For a charity with 40,000 donors not currently under Gift Aid declarations, a call programme converting 25% to valid declarations adds £300,000 in annual income, at an AI voice calling cost of £10,000-15,000. The same programme run by a human calling agency costs £240,000-800,000.

McKinsey context: The 2025 State of AI report found that 88% of enterprises now use AI in at least one function, but only 6% capture material efficiency gains from deployment. Charity fundraising AI voice sits in the 6% when the compliance architecture is built in from the start — the programmes that fail do so because PECR breaches trigger enforcement events that close the programme down.

---

What to look for in an AI voice platform for charity fundraising

Not every AI voice platform is built for the compliance requirements of UK charity outbound calling. When evaluating a platform, these are the questions to ask before any commercial commitment:

PECR and TPS compliance

• Does the platform support pre-campaign TPS and CTPS suppression, with a downloadable audit log?
• Can it integrate FPS API or batch-file checks before each campaign run?
• Does it maintain an internal DNC list with real-time suppression and propagation to downstream campaigns?

Fundraising Regulator Code compliance

• Can the opening disclosure script be locked at the system prompt level — not editable by individual campaign managers — to ensure consistent AI disclosure on every call?
• Does the platform support CLI presentation of a real, contactable number for every outbound call?
• Does it generate the evidence file the Fundraising Regulator would request on audit?

Opt-out processing

• Does the platform recognise natural-language opt-out intent, not just keyword triggers?
• Does it write opt-out flags to the charity's CRM in real time (not batch)?
• Does it generate a suppression confirmation for the donor record?

Consent capture and Gift Aid

• Can the platform capture verbal consent with a timestamp and write it to the CRM before the call ends?
• Is the consent record in a format suitable for HMRC Gift Aid audit — including the audio record, timestamp, and donor identifier?
• How long are consent records retained, and can retention periods be configured to the charity's own data retention policy?

Audit and ICO-readiness

• Can the platform export suppression logs, call records, consent captures, and opt-out records in a format suitable for ICO investigation response?
• What is the retention period for call recordings, and can it be aligned with the charity's GDPR data retention schedule?

Ready to model the programme economics for your donor base? Try Dilr Voice live (free, £20 credits), review the AI placement diagnostic for third-sector programme design, or explore our approach to regulated-sector AI voice deployment.

---

Regulatory references

The regulatory framework for charity telephone fundraising is not static. The key sources to monitor:

• ICO PECR guidance on live calls — updated periodically; the ICO's direct marketing guidance covers charity callers explicitly: ICO PECR telephone marketing guidance
• Fundraising Regulator Code of Fundraising Practice, Section 11 — the governing standard for telephone fundraising practice: Fundraising Regulator Code
• Fundraising Preference Service — the sector-specific opt-out register every charity calling programme must check
• TPS registration and scrubbing service — PECR-required suppression for all outbound live calls

---

Written by the Dilr.ai engineering team — practitioners who ship enterprise AI voice in regulated sectors. Follow us on LinkedIn for deployment notes, or subscribe via the RSS feed.