Nearly half of all press releases distributed through PR Newswire in June contained AI-generated text, according to a new analysis that scanned 178,672 English-language releases over a 12-month period. The findings, from AI detection firm Pangram, signal a rapid shift in how companies handle public relations writing - one that carries implications for message quality, trustworthiness, and the sheer volume of corporate communications flooding newsrooms.
The study found that 34% of releases published on the wire service between August 2025 and July 2026 flagged for AI content. The weekly share doubled over that year, climbing from 22% to a peak of 48% in June 2026. Of the 61,523 releases that contained AI-generated text, 25% were fully AI-authored, while the remaining 75% mixed human and machine writing.
The numbers behind the shift
AI contribution was substantial when it appeared. The median release flagged for AI content was 52% machine-written. Across all flagged documents, the distribution was bimodal - showing clusters of light AI editing on one end and full AI generation on the other.
Word counts are creeping up alongside AI adoption. The median press release grew from 488 to 562 words over the twelve months studied. In July 2026, releases with AI-authored text averaged 601 words, compared to 521 words for fully human-written ones. AI appears to be enabling longer, if not necessarily more substantive, communications.
Pangram also rescanned a previous academic dataset from Liang et al. to track adoption over a longer timeline. The combined results show AI uptake took two years to reach 9% by August 2024, then doubled to 22% by July 2025, and doubled again to 44% by July 2026. Contrary to earlier estimates, adoption was gradual in the months after ChatGPT's launch - Pangram found it took nearly a year for press releases to break a 5% AI contribution rate.
The detection method proved precise on pre-ChatGPT releases. When tested against 21,625 releases published before November 2022, Pangram's tool returned a single false positive, a false positive rate of 0.0046%.
Which industries lead and which lag
Advertising topped the list, with 64% of releases containing AI-generated text and 41% flagged as fully AI-written. Much of that volume traces back to content marketing agencies producing high-volume editorial-style articles for local businesses. One issuer, HelloNation, accounted for 2,265 releases in the dataset - 97% of them fully AI-generated, with just one flagged as human-written.
Finance, energy, and biotechnology sat at the opposite end of the spectrum. Each had AI contribution rates below 5% for fully AI-authored releases. The pattern held across sectors: industries dominated by small businesses and startups showed far higher AI adoption than those controlled by large, established institutions. Financial technology releases, often from single-person startups and crypto ventures, had a 44% AI contribution rate with 12% fully AI-generated. Traditional banking and financial services sat at 25%, with only 7% fully AI-generated.
The same divide showed up when comparing wire services. Budget distributors PRWeb and Newswire.com, which cater to small and midsized businesses, both had AI contribution rates close to double that of PR Newswire.
What's at stake for communications professionals
None of the wire services studied require AI disclosure. PR Newswire sells its own AI writing tool and states plainly that it "does not publicly tag press releases to indicate they were created using AI." The absence of labeling means journalists and investors cannot distinguish machine-written material from human-authored releases.
The quality concerns are real. AI-generated text can introduce factual errors - a serious liability for releases tied to SEC disclosure obligations. Readers also find AI writing less trustworthy, and fatigue around common large language model phrases continues to grow. Yet the polished, grammatical, if formulaic, output of AI tools mirrors the same criticisms often leveled at human-authored corporate prose, making the machine fingerprints easier to overlook.
Volume may be the more disruptive force. AI enables anyone to produce press releases at scale for little more than the cost of distribution. The resulting flood of low-information text risks degrading the signal across professional communication channels. Smaller players - including genuine local businesses competing with AI-generated content mills - stand to lose the most visibility in an increasingly noisy environment. For PR and communications teams, the challenge ahead involves not just deciding whether to use these tools, but navigating a landscape where every competitor, client, and counterparty can fill the pipe faster than ever before.
Why this matters for PR and communications professionals
The press release remains a foundational tool for corporate disclosure, media relations, and brand positioning. When nearly half of releases on the largest U.S. syndication service contain AI-generated text - and that share is doubling annually - the baseline expectations for what constitutes original, trustworthy corporate communication are shifting in real time. PR professionals who understand how AI detection works, where their own industry sits on the adoption curve, and what disclosure norms are emerging will be better positioned to advise clients and employers on when machine-generated text helps and when it harms. For those looking to build those skills, AI for PR & Communications resources and AI Courses for Media Relations Specialists offer practical grounding in the tools reshaping the field.
Your membership also unlocks:
