How AI Will Change PR and Communications in 2027, and How Agencies Can Prepare: A TNS Study

76% of PR professionals use AI; the survey says use has levelled off

Muck Rack, which sells PR software, says generative AI use in PR "has peaked at 76%" (564 PR professionals, 5 to 24 December 2025). It warns that changed question wording makes year on year comparisons weak.

93% say AI makes them faster. The gains are self-reported

93% of the survey's AI users say AI makes them faster, and 82% say it improves quality; 98% always or often edit AI text.

Every one of these figures is self-reported. We found no study that compares PR results with and without AI.

Only 51% have an employer AI policy, and 43% get training

Only 51% of PR professionals have an employer AI policy and 43% get training, both up from 21% in 2024 (the same December survey). And while 77% of the survey's AI users avoid entering financial data, only 36% avoid entering client names.

82% of journalists use AI; in another survey, 53% oppose AI-generated pitches

82% of journalists use at least one AI tool, up from 77% a year earlier (the same firm's State of Journalism 2026: 897 responses analysed, 30 January to 2 March 2026, 81% in North America). Cision's release on its 2026 State of the Media (1,899 journalists in 19 markets across North America, EMEA and APAC, January and February 2026) says "53% oppose AI-generated pitches due to concerns around accuracy and personalization." Both firms sell PR software.

Opposition is not universal. Cision's own article on the survey puts those in favour at 21%, and its chief executive later wrote in Press Gazette that 25% were neutral.

In the mostly North American State of Journalism survey, journalists delete pitches that miss their beat (88%), are overly promotional (71%) or arrive as mass email (50%).

Labour-based agency pay fell from 54% of multinationals in 2011 to 19% today

Among 69 multinationals with USD 147 billion of marketing spend, labour-based agency pay fell from 54% in 2011 and 33% in 2022 to 19% today, while fixed fee or output models rose from 20% to 33%. The survey covers advertising and marketing agencies, not PR only, and 71% of respondents work in procurement.

Omnicom cut jobs, but its CFO says AI is "not necessarily the primary driver"

Omnicom announced about 4,000 job cuts after buying IPG (Axios, 2 December 2025). On its results call of 18 February 2026, its chief financial officer said: "AI is not necessarily the primary driver of how we looked at this." On the same call, its chief executive said AI-enabled automation will let it eliminate some positions that are now done by hand.

Only 3% of agencies report significant staff cuts due to AI

77% of PR professionals fear the next generation will not learn the basics (Muck Rack). But only 3% of agencies report significant staff reductions due to AI (Productive.io, which sells agency software; 174 agencies and consultancies that use Productive, March 2026, 54% in Europe, not PR specific).

86% want AI content disclosed, but 32% would trust a brand less if its content is AI-made

86% of consumers say AI-generated content should be disclosed, yet 32% say they would trust a brand less if its content is AI-generated, and 15% would trust it more. That is the disclosure paradox: people want to be told, and being told can cost trust.

The survey, by Meltwater, which sells media monitoring tools, with YouGov, covered nearly 10,000 consumers in seven markets (21 April 2026); a news report on the study lists none in Latin America.

Only 18% of corporate affairs functions say they are ready for a deepfake

In GlobeScan's survey of nearly 300 senior corporate affairs practitioners, 18% say their function is prepared for a deepfake or AI misinformation incident. Respondents in Latin America feel the most prepared, which GlobeScan says may reflect "an optimistic underestimation of the threat ahead."

EU disclosure rules have applied since 2 August 2026

The EU AI Act's transparency duties in Article 50 apply from 2 August 2026. Deepfakes must be disclosed, and so must AI-generated text published to inform the public on matters of public interest, unless a human has reviewed it and someone holds editorial responsibility for it.

This is our reading, not legal advice.

13% of Brazil's online sample uses AI chatbots for news weekly, against 10% globally

In Brazil's online sample, 13% use AI chatbots for news each week, above the 10% global average, and 19% of under-25s do; Peru is at 11%, Argentina and Colombia at 9% (Reuters Institute, Digital News Report 2026, online samples of about 2,000 per market).

In Buenos Aires, 91% of communicators use AI, but 64% got no training

In the Buenos Aires area, 91% of 318 communicators use AI at work, but 64% got no training from their employer (UADE, fieldwork November 2025, non-probability sample).

Brazil's election: only 371 of 920 political-electoral posts confirmed as AI carried a label

Under Brazil's electoral court rules for the 2026 election (TSE Resolution 23.755, 2 March 2026), AI-made electoral advertising had to carry a label that is "explícito, destacado e acessível" (explicit, prominent and accessible).

From 16 August to 28 September 2026, Lupa's VigIA monitor, with Unicamp's Recod.ai lab, confirmed AI in 920 political-electoral posts, including at least 554 deepfakes; only 371 of the 920 carried a label (as reported by Agência Brasil, 2 October 2026; we did not find Lupa's own report).

Only Peru, among TNS markets, has a general AI law in force

Sources: official records where we found them, otherwise the outlets named in the references. Checked 10 November 2026; not legal advice.

How we built this study

Questions and answers

References

Each source is listed once, under the first chapter (01 to 06) that cites it, or under Chapter 00 if only the opening cites it.