|
|
Was this forwarded to you? Get The Fourth Estate AI Brief in your inbox →
|
|
|
|
View this issue in your browser
Tuesday Brief
The Bot Lobby, The Meta Dinner, and Your AI Budget
The interesting thing this week with AI and publishing: the battle lines are getting clearer. Three hundred publishers flew to D.C. to fight bots. One Louisiana owner hosted a dinner for Meta hoping for a licensing deal. And Anthropic just cut AI costs by 40%. These aren't disconnected stories. They're three different bets on the same question: who controls the value of your content, and what's it worth?
|
|
PUBLISHERS VS. THE BOTS
The Rundown: The News/Media Alliance doubled last year's D.C. fly-in, bringing CEOs from Condé Nast, Hearst, USA Today, and The Seattle Times to push the Stealth Bot Prohibition Act.
The details:
- The bill would require AI crawlers to identify themselves and disclose their purpose, so no more hiding behind residential IP proxies.
- Bot traffic now exceeds human traffic on the internet. Cloudflare data shows bots make up over 60% of all web visits.
- TollBit detected more than 22 billion AI bot scrapes in the first half of 2026 across sites it analyzed.
- AI bot visits to publisher sites surged sixfold in 2025, from 1 in 200 human visits in Q1 to 1 in 31 by Q4.
- A similar state bill already passed in New York in June; the federal version was introduced in July.
Why it matters for us: If your site has been sluggish or your server bills have spiked unexpectedly, stealth bots may be the cause. This bill would let you actually see and block them, giving you a leg up before any licensing conversation starts.
Read at digiday.com →
|
|
|
THE META DINNER
The Rundown: John Georges, owner of The Times-Picayune, co-hosted a celebratory dinner for Meta alongside the governor, hoping for a licensing deal like News Corp's $150 million contract.
The details:
- Georges Media Group's logo appeared alongside Meta's on invitations for a dinner celebrating Meta's $50 billion Louisiana data center.
- At least one staff column critical of the company's relationship with Meta was killed by management.
- The Times-Picayune ran sponsored content videos on the data center with 'sponsored by Meta' labels that faded after a few seconds.
- The Times-Picayune has lost two-thirds of peak staff, down to roughly 450 employees.
- Sunday circulation has dropped 25% since 2015.
Why it matters for us: This is what happens when business development interests collide with editorial independence. The line between advertising partnerships and coverage gets blurry fast, and staff trust erodes when columns get killed.
Read at wwno.org →
|
|
|
ANTHROPIC'S PRICE CUT
The Rundown: Anthropic released its new flagship model with a significant price drop: $4 per million input tokens, $20 per million output tokens, and 30% faster generation.
The details:
- Opus 5.5 costs 40% less than Opus 5 on typical workloads, with cache reads at $0.20 per million tokens.
- One tester completed a 680,000-line code migration in less than a day, work that would have taken weeks.
- The model generates output 30% faster than Opus 5 and uses fewer tokens per task.
- Anthropic increased usage limits on Pro, Max, Team, and Enterprise plans.
- Claude Sonnet 5.5 and Claude Haiku 5.5 will follow in the coming weeks.
Why it matters for us: If your publication uses Claude for transcription cleanup, draft generation, or document analysis, the 40% cost drop means you can run more tasks on the same budget, or cut your monthly AI spend while keeping output steady.
Read at anthropic.com →
|
|
|
THE ENTRY-LEVEL SQUEEZE
The Rundown: AI handles the tasks that used to train tomorrow's editors.
The details:
- McClatchy Media laid off 30% of workers across its 17 publications this month, replacing them with AI tools.
- The Associated Press now uses generative AI for headlines, data analysis, and video production support.
- A journalism CEO told students that AI is offloading entry-level jobs, and 'anyone telling you otherwise is lying.'
- A senior lecturer argues AI frees journalists from 'menial tasks' for on-the-ground reporting.
Why it matters for us: If your publication relies on interns or early-career reporters to grind through meeting minutes and police blotter, you're training the people who become your managing editors in five years. Cut that rung and you've got a ladder missing its bottom step.
Read at dailycollegian.com →
|
|
|
THE UPDATE
Quick Hits From the Week
The Rundown: Small items that shipped, leaked, or surfaced, but did not justify a full section.
The details:
- New York's state-level bot disclosure law goes into effect October 1, making it the first state with enforceable AI crawler rules.
- OpenAI quietly updated its robots.txt guidance, acknowledging that some publishers have seen crawl rates drop 80% after blocking GPTBot.
- Google's NotebookLM added a 'source attribution' toggle that lets users see which uploaded documents generated each summary line.
Why it matters for us: Worth watching but not worth a full breakdown this week.
|
|
|
What is?
Bot
What it is: A bot is a program that visits websites on its own, with nobody at the keyboard. Most traffic on the internet is already bots, and they are not all the same animal. Three kinds matter to you now. Training crawlers copy your stories to help build AI models and send you nothing back: GPTBot, ClaudeBot, Google-Extended, and CCBot are the common names. Answer crawlers fetch your page so a chatbot can quote it, usually with a link back to you: OAI-SearchBot, Claude-SearchBot, and PerplexityBot. And agent bots are running an errand for one real person who asked their AI to go look something up. Same word, three very different deals for you.
Why publishers care: Because the word covers both the bot copying your archive and the bot sending you a reader, and the answer is different for each. Your robots.txt file is a request, not a lock. The well-behaved crawlers honor it, and last year Cloudflare caught one AI company's crawler dodging blocks by disguising itself as an ordinary Chrome browser and coming in through home internet addresses. That is the behavior the bill in today's lead story goes after. The practical version for your own site: block the training crawlers, allow the answer crawlers, and know that blocking everything also takes you out of the AI answers more of your readers are using instead of search. If your site runs through Cloudflare, it now gives you separate switches for search bots, training bots, and agent bots instead of one on-off, at no cost. Find out who is knocking before you decide what to do about it.
|
|
|
|
Three hundred publishers flew to Washington. One hosted Meta for dinner. And the cost of running AI just dropped 40%. The question isn't whether AI changes the economics of your publication. It's whether you're shaping those economics or just reacting to them.
Trevor Slette runs Quadd.ai, AI tools built for publishers.
|
|
|
|
|
|
Run a newspaper. Use Quadd.
AI tools built for small newsrooms. Document intelligence that turns court reports and box scores into copy you can paste into your layout. Audio transcription that finds the quote without you scrubbing for it. AP-style proofreading that tightens copy without flattening voice. Built by someone who knows what a Tuesday night feels like. Free for seven days.
Request a 7-day trial →
|
|
|
