What happened
Cision's 2026 State of the Media report, based on 1,899 journalist responses across North America, EMEA, and APAC surveyed in January and February 2026, finds 53% of journalists opposed to AI-generated pitches on grounds of accuracy and personalization. The share of journalists naming AI as one of the top industry challenges climbed to 43%, up from 30% in the 2025 report. At the same time, journalists' own AI usage is up sharply — 82% now use at least one AI tool according to Muck Rack's parallel 2026 State of Journalism data — creating a widening asymmetry: journalists rely on AI to work faster but reject AI when it shows up in their inbox. Cision's guidance in the report is to use AI on the input side (fact-cleaning, angle tailoring, asset packaging) rather than the output side (pitch drafting). PR software vendors including Prowly and Muck Rack are updating pitch-composer features to reflect the input-versus-output split.
Why AI-generated PR pitches now fail
MagnetizeX builds founder visibility systems for B2B firms.
The 53% opposition threshold flips the previous PR-tool math. A year ago, an AI-drafted pitch could sneak through if it was tailored enough. Now, more than half of journalists are pre-screening for AI patterns — em-dashes, generic openings, made-up hooks — and disregarding on first read. AI-generated PR pitches consistent with common LLM output patterns are dead on arrival at over half the outlets B2B founders want. What still works is the input-layer use: AI to research the reporter's last five stories, cross-reference angles, and prepare assets. The pitch itself now has to sound like a person wrote it, because it has to survive a filter set to look for the opposite.
- Write the pitch first, edit with AI second
Draft every pitch by hand in under 120 words, then run it through Claude or ChatGPT with the prompt "flag any phrases that sound AI-generated." Rewrite those specific phrases. This inverts the typical workflow and keeps the pitch on the human side of the 53% opposition line.
- Use AI for reporter research, not outreach
Build a Claude or ChatGPT prompt that ingests the reporter's last ten bylines and outputs three specific angle recommendations tied to your news. The input-layer use is what Cision itself endorses, and it lifts pitch relevance without triggering the AI-detection instinct that kills the pitch.
- Kill AI subject lines this week
Subject-line generators are the loudest AI signal in a journalist's inbox. Write subject lines by hand this week, cap at 8 words, name the specific outlet-relevant angle. Muck Rack data shows 88% of journalists disregard off-beat pitches on first read, and generic AI subject lines pattern-match to off-beat.
By the numbers: 53% of journalists oppose AI-generated pitches, 43% cite AI as a top industry challenge (up from 30%), 82% of journalists use at least one AI tool themselves, and 88% disregard pitches that miss their beat on first read.
What to do this week
Pull your last five outbound PR pitches and score them for AI-tell density using a Claude prompt: paste the pitch, ask "which phrases here read as AI-generated." Rewrite each flagged phrase in first-person voice referencing a specific detail from the reporter's recent coverage. Prowly's pitch composer and Muck Rack's newer AI features both support the input-layer workflow now. This is a 30-minute exercise per pitch that materially lifts response rate on the Cision-tracked outlets B2B founders actually care about.