Media Mention: Yes, AI Can Write Your Pitch. But Can It Give Journalists What They Need?

Highlights

• October 15, 2026
• 79% · of HARO pitches had a human involved · 3x · higher feature rate for human-written pitches · ~2/3 · of classified AI citations trace to editorial sources
• AI has found its way into media pitching, but a new report from Help a Reporter Out (HARO) suggests most pitches still have a human behind them. Of more than 2 million pitches HARO analyzed, 79 percent showed some human involvement, while 21 percent -- roughly 420,000 pitches -- came back as fully AI-generated, HARO CEO Brett Farmiloe told EIN Presswire. Human-written pitches were about three times more likely to get featured than entirely AI-written ones.
• Key findings from HARO's pitch analysis
• Human pitches win about 3x more often: Human-written pitches were roughly three times more likely to get featured than fully AI-generated ones, according to HARO's analysis of more than 2 million pitches.
• 21% of pitches are fully AI-written: About 420,000 of the pitches HARO analyzed were entirely AI-generated, with no human involvement in the writing process.
• 35% of journalists block AI pitches: More than a third of journalists have switched on HARO's filter that hides pitches flagged as AI-generated. The filter is unchecked by default.
• Earned media feeds AI answers: Featured's AI Citation Report found editorial sources made up roughly two-thirds of classified citations across 22,881 AI citations, while brand-owned content accounted for just 1.7 percent.
• Most journalists I speak with say they can tell when a pitch is AI generated or not, just from a 'blink' test -- in one second or less, they can just tell.
• Farmiloe's advice to publicists doubles as the HARO acronym itself: be Helpful, be Authentic, be Relevant, be On time. Helpful means the pitch genuinely helps the journalist do their job. Authentic means the final pitch goes beyond what a generic AI model would produce on its own. Relevant means the source's credentials actually match the request. And 'on time' isn't about being first -- it's about arriving inside the journalist's working window with the best credentials and the most useful answer. A typical query draws about 42 pitches, Farmiloe said, and the math was never friendly to volume.
• The interview also touched on Featured's AI Citation Report, published in August. The study ran 405 AI visibility audits across 378 brands, asked AI platforms 1,983 buying-style questions, and traced 22,881 resulting citations back to 11,499 domains. The median cited domain had a domain authority of 37 -- 31.2 percent of citations went to sites with a domain authority of 80 or higher, while 34.5 percent went to sites under 40. 'Media mentions help with model mentions,' Farmiloe said.
• AI can polish any pitch into an authoritative stance. Now, it matters more about who is sharing that stance.
• HARO runs every pitch through AI detection with partner Pangram and reviews every journalist request before it goes out, requiring a minimum Moz domain authority of 20 for website outlets (podcasts are exempt); about 32 percent of requests get rejected at that step. Source profiles pass seven or eight verification checks before reaching a journalist, and HARO bans an account roughly every 23 minutes, mostly for spam. An AI score alone, Farmiloe noted, is never grounds for a ban -- it's one signal among many.