Anthropic put Claude Opus 5 on sale today at $5 per million input tokens and $25 per million output tokens, the exact prices it charged for the previous Opus 4.8. According to Anthropic’s announcement, the model reaches close to the intelligence of its higher-end Fable 5 model at half the cost, and it now serves as the default model on Claude Max and the strongest option on Claude Pro.

The company is leaning hard on benchmark numbers to make its case. On Frontier-Bench v0.1, Anthropic says Opus 5 beats every other model and more than doubles Opus 4.8’s score at a lower cost per task. On ARC-AGI, a test built around novel reasoning problems, the company reports Opus 5 scoring three times higher than the next-best model. On Zapier’s AutomationBench, which checks whether a model can run a business task end to end, Anthropic claims a pass rate roughly 1.5 times the nearest competitor at the same cost.

The recurring theme in the launch is verification. Anthropic describes Opus 5 checking its own work before handing it back. In one Frontier-Bench task, the model was asked to rebuild a machine part in 3D code but given no way to view the drawing, so it wrote its own computer vision pipeline to pull the geometry from raw pixels. In another example, it found the root cause of a bug in an open-source package manager that the community’s own patch had missed.

Early-access customers echoed that framing. JetBrains said the model catches its own logical faults during planning rather than after. A legal-tech tester reported first-turn redline scores nearly double Opus 4.8. Box measured an 8% overall improvement over Opus 4.8, with 17% gains on due-diligence workflows. These are vendor-supplied quotes, so treat the precise figures as marketing rather than independent measurement.

On safety, Anthropic says Opus 5 is its most aligned model so far, scoring 2.3 on its automated misaligned-behavior audit, the lowest of its recent releases. The company also notes the model stays behind its Mythos 5 model on biology research and offensive cybersecurity. Notably, Opus 5 can find software vulnerabilities about as well as Mythos 5 but lags badly at writing exploits for them, which Anthropic frames as a deliberate safeguard. Its cyber classifiers block binary vulnerability scanning, penetration testing and exploit generation, with flagged requests falling back to Opus 4.8.

For newsrooms and publishers, the pricing is the story. Holding costs flat while claiming stronger reasoning and cleaner outputs matters for teams running document analysis, research summaries and data work at volume. The customer notes about tighter, more concise responses and fewer tool calls point to lower token spend per task, which is where AI budgets actually get decided. Publishers weighing model choices should still run their own tests against real workflows rather than trusting benchmark charts, a point we’ve made repeatedly at The Media Copilot.

Anthropic paired the launch with two beta features: mid-conversation tool changes that don’t break the prompt cache, and automatic API fallbacks that route flagged requests to another model instead of blocking them. A Fast mode runs about 2.5 times the default speed at twice the base price. The bet is that a cheaper, more careful default beats a smarter but pricier one for daily use, and the next few months of real deployments will show whether that holds.

The post Anthropic ships Claude Opus 5, pitching frontier work at half the price appeared first on The Media Copilot.

As AI companies continue scraping the web for training data, some publishers are experimenting with a new tactic that doesn’t only block unwanted bots, but tries to waste their time and money.

Known as LLM honeypotting, Digiday describes the approach as a form of deception to lure AI crawlers into consuming plausible-looking but ultimately worthless content, aimed at making large-scale data collecting so computationally expensive that it becomes less economically viable. 

Who’s doing this? A small number of publishers and e-commerce companies are looking for alternatives to traditional bot-blocking. And while the technique remains early and experimental, it’s one way for media companies to protect their content amid a fight to create standards for how AI companies track, value and compensate journalism. 

A cybersecurity tactic adapted for AI

Cyberhoneypotting is a long-standing cybersecurity strategy. The technique is to build a decoy target for attackers—or LLM bots, in this case—that can lure bad actors away and even gather intelligence on their capabilities and methods. In this case, the goal is to make scraping content without compensation more costly than it’s worth. 

Simon Wistow, co-founder of content delivery network provider Fastly, describes the philosophy as one of “Chang[ing] the economics of attacking.” If abusing a system becomes significantly more expensive than the value gained, the entire model will become unsustainable, he argues. 

Applied to AI crawlers, the strategy is equipped against all automated visitors, regardless of whether they are operated by major AI companies or smaller third-party scraping firms. 

Publishers can implement the tactic in several ways. They can introduce subtle delays, difficult (for computers) problems to solve before admittance, or endless mazes of contents and files filled with meaningless AI-generated text. Some honeypotting techniques go even further and try to inject bad data into the bots’ training datasets, “poisoning” the AI results.

The tactic’s adoption

Large e-commerce brands are already testing the technology successfully, said Wistow. News publishers are also showing increased interest, although he declined to identify specific customers. 

Even so, skepticism towards the strategy remains. 

Frederick Jahn, co-founder of AI company Centennal, argues sophisticated scrapers can often detect or avoid honeypots altogether. 

“I think it’s a good concept, but more on a marketing level, and like a gimmick,” Jahn said. He argues that publishers would be better served by creating real barriers to stealth crawlers, who are often not shown maze pages. 

Supporters of honeypotting maintain that, even if most scrapers adapt, increasing operational costs across thousands or millions of requests could make smaller scraping businesses financially unsustainable. 

“If they could burn through that 10 million funding in one crawl then suddenly those businesses aren’t viable and suddenly the whole market collapses, and that’s kind of what you’re going for,” said Wistow.

Costs and limitations for publishers

The strategy isn’t free. Generating and serving millions of fake pages is more expensive than simply blocking unwanted traffic. And larger publishers with more resources are better able to plan and implement the strategy.

Wistow said the approach is unlikely to become widespread, in part because of the consequences of filling the internet with yet more intentionally deceptive content. 

“Hallucinations happen even with good data, just because of the way LLMs work,” said Wistow. “This is about changing the economics for the people abusing your site, not running some giant disinformation campaign.”

The post Publishers Turn to AI ‘Honeypots’ to Fight Content Scraping appeared first on The Media Copilot.

Organic Google Search referrals to eight major UK publishing groups fell 7.1% between the first and second quarters of 2025. If that holds, the volume of pageviews Google sends to these publishers will be cut in half by the third quarter of 2027.

That projection comes from the Association of Online Publishers’ new Artificial Intelligence Publisher Impact Study, detailed by Richard Reeves in Press Gazette. The AOP commissioned DJB Strategies to analyze six months of traffic across 10.8 billion pageviews, roughly 80% of the participants’ total after long-tail URLs were stripped out, alongside an Ipsos survey of consumer AI use.

The study revealed the decline even as Google reported that its Search revenue grew 12% and expanded AI Overviews, as the company turns Search from a route to content into a destination that answers queries directly. With Google pushing its zero-click AI Mode toward the center of the search experience, the study’s authors expect the drop to speed up rather than level off.

News publishers held up better than most, averaging a quarterly decline of around 5%, against 18% for the consumer and B2B segment. The study attributes that to tighter guardrails on news inside AI Overviews compared with the evergreen content that AI aggregates easily. Split by topic, news actually ticked up 2%. Sports, travel and health & wellness gained 37%, 22% and 7% respectively, while technology, finance & business and politics fared worst, falling 28%, 23% and 7%.

Guide and how-to content lost more than 20% of its Google referrals, exactly the kind of query a reader gets from an AI answer instead of clicking through. Content carrying an explicit point of view resisted substitution best. Opinion and commentary referrals rose 4%, interviews rose 1%, and product reviews stayed roughly flat.

The catch is that reportage, the largest single content category, accounted for more traffic than everything else combined and still slipped 4%. The study suggests finding ways to inject reportage with a more distinctly human voice, which lines up with the broader argument that authority is becoming the new speed as generic information retrieval moves to the machines.

Sites funded largely by high-volume programmatic advertising rely the most on transient search traffic and have the least cushion. Reader-funded operations, however, can absorb the shock.

The study’s warning is blunt: don’t chase falling traffic by pumping out more URLs in the hardest-hit categories. That only dilutes views-per-page and accelerates the losses. The volume game built on basic information retrieval is ending, and Google is the one turning out the lights. The publishers who survive will be the ones producing work an AI answer can’t cleanly replace.

The post Google search traffic to drop by half by Q3 2027 for UK publishers appeared first on The Media Copilot.

Layoffs in the U.S. and U.K. journalism industry topped 2,300 newsroom jobs in the first half of 2026, Press Gazette reported July 14, a trend that, if it continues, represents an almost 34% increase over all of 2025’s journalism job losses. The majority of jobs lost are among public broadcasters and wire services alongside newspapers, magazines and digital publishers.

The Press Gazette’s running tally, tracked at least 3,434 jobs lost in 2025 and at least 3,875 in 2024.

Publishers increasingly describe the changes as newsroom reorganizations rather than cost-cutting measures, the underlying pressures remain visible. As audiences migrate to video, social platforms and AI-powered search and chat interfaces, news organizations are redirecting resources toward digital products, audience growth and new distribution strategies. 

The BBC has said it plans to eliminate up to 2,000 positions, roughly one in 10 employees, in what would be the broadcaster’s largest workforce reduction in 15 years. BBC News is expected to bear much of the impact as the organization seeks hundreds of millions of pounds in savings. 

In the U.S., NPR is cutting up to 30 newsroom positions while offering buyouts to hundreds of employees as it meets shifting audience habits and the loss of federal subsidies. The Associated Press is also reducing its editorial workforce through a combination of buyouts and layoffs as it shifts investment toward visual journalism and digital products. 

Other publishers have announced smaller but persistent reductions. CBS News cut about 66 employees and has shut down its century-old radio division. The Washington Post is shrinking parts of its newsroom and business operations, while Vox Media, Condé Nast, Future, Politico, Bustle Digital Group, The Standard, and Nexstar Media Group have also reduced their editorial staff this year. 

At the same time, many are expanding their use of AI to automate production workflows, personalize recommendations and support reporting, raising questions about the fate of the traditional newsroom structure as emerging technologies reshape the industry.   

The post Journalism’s workforce shrinks as AI and new consumer habits reshape the industry appeared first on The Media Copilot.

Germany has become the first country to explicitly classify AI search features, such as Google’s AI Overviews and Perplexity’s answers, as media products subject to national media law, according to Reuters. 

The country’s Commission for Licensing and Supervision, which represents Germany’s 14 state media authorities, has argued that AI summaries and chatbot answers go beyond aggregating existing information as they create new outputs that reflect the providers’ own systems. 

This follows a separate June 2026 court case in which a Munich court found Google could be held directly liable for inaccurate information produced by its AI Overview feature, according to the German newspaper publishers’ association BDZV.

Germany’s media regulator said AI search tools fall outside the European Union’s usual platform liability protections because they are not merely hosting third-party content but curating their own responses. The regulator also said Google’s AI Overviews and Perplexity’s chatbot answers can shape what news audiences see, raising concerns about publishers’ visibility and reach.  

As a result, the services provided by AI search features are therefore subject to German regulations aimed at protecting media plurality.

The decision could give regulators and publishers worldwide a new foothold in their push to hold AI companies accountable for how generated answers reshape the news ecosystem. It also weakens one of the key defenses used by technology platforms, which claim only to be intermediaries organizing information created by others. 

The case may also become a blueprint for other European regulators, which have intensified investigations into major technology companies over competition, privacy and online safety concerns in recent years. 

Google intends to appeal the decision, with a spokesperson characterizing the classification as a failure “to recognize how people’s preferences when searching for information and the information ecosystem are changing.”

The post Germany says AI Overviews subject to national media law appeared first on The Media Copilot.

The growing popularity of AI chatbots is reshaping how audiences discover news, with one in 10 people worldwide now using tools such as ChatGPT, Perplexity and Google’s AI Overviews for news at least weekly, according to the Reuters Institute for the Study of Journalism’s Digital News Report 2026. 

On July 13, Arc XP, a media operating system developed by the Washington Post, launched Ask The News, an AI-powered conversational search tool that publishers can embed on their own sites. Users can ask questions of Ask The News, which answers using the publisher’s reporting with attribution and editorial safeguards. The tool will decline to answer questions when it does not have enough source reporting to generate a confident response, Arc XP says, rather than hallucinate an answer. 

Arc XP’s Subscription Gateway, a capability embedded into Ask The News, changes the traditional paywall model by limiting AI-generated answers rather than articles, prompting readers to subscribe after they have received value from the tool.

For newsrooms, the tool could offer a new way to understand audience demand. Instead of relying on clicks and page views, Ask The News allows publishers to see what questions readers have and subsequently use those insights to guide coverage decisions. 

“Search shows links. Generic AI gives answers from the open web,” said Joey Marburger, Vice President of Content Intelligence, Arc XP. “Ask The News answers with a publisher’s journalism, rules, and business model. This is about building a durable reader habit: when you have a question, you go to your trusted news organization first.”

As AI-generated answers become an increasingly common way for people to access information, publishers may be forced to rethink how their reporting reaches audiences through emerging AI experiences in addition to through traditional means. 

Ask The News is now available to Arc XP publisher partners with a JavaScript embed option for non-Arc XP publishers. Arc XP plans to expand the tool with features including personalized briefings, saved conversations, topic tracking and editorial intelligence dashboards. 

The post The Washington Post’s Arc XP launches AI answer tool to help keep readers in their ecosystems appeared first on The Media Copilot.

In 2023, a company called Alabama Community News LLC supposedly spent $3.2 million to buy 47 weekly newspapers across the state. The corporate owners fired the local staff, replaced them with an artificial intelligence system that scraped high school sports scores, and promptly drove the entire network into bankruptcy. The story even named a specific 26-year-old campaign staffer who generated 70 percent of the copy.

None of it actually happened. The entire 1,900-word saga was a fabrication published by a site called The Editorial, according to an investigation by Nieman Lab. The targeted newspapers, including the Shelby County Reporter and the Centreville Press, are still printing. The angry local advertisers quoted in the piece do not exist. The story falsely claimed the roll-up was funded by 1819 News, a real conservative outlet in the state, adding a layer of plausibility to the hoax.

The fake story gained traction among journalists on social media platforms like Bluesky before the operators pulled it down. They replaced the page with a sterile retraction notice citing “fact-verification concerns.” But the Alabama hoax was not an isolated incident.

The Editorial has built a bizarre subgenre of AI-generated obituaries for real American newspapers. The site previously published fabricated stories detailing the collapse of the Chattanooga Times Free Press, the Kenosha News, and the Macon Telegraph. The nonexistent reporters credited with these stories sport fake resumes claiming past stints at ProPublica and Reuters.

The motive behind the site remains murky. Domain registration and payment records point to a Finnish technology company called Nordiso Group, which develops AI study apps. Yet the site’s political sections suggest a different angle. The Editorial publishes a high volume of geopolitical content focused on Taiwan and the South China Sea, heavily pushing narratives that highlight Chinese military dominance.

These geopolitical stories share obvious synthetic fingerprints. Nearly every piece opens with a variation of the exact same scene: a nondescript, windowless conference room where a secret document slides across a table. This repetitive structure aligns with tactics tracked by groups like the Stanford Internet Observatory, which monitors state-sponsored disinformation campaigns. It also highlights how cheap synthetic media allows operators to flood niche topics, a trend we track closely at The Media Copilot.

For publishers, this represents a strange new vector of reputational risk. Newsrooms are used to fighting disinformation about elections or public health, but now they must monitor for synthetic hoaxes about their own business operations. A fake story about a newspaper shutting down or firing its staff can spook actual advertisers and confuse real subscribers before the publisher even realizes the rumor exists.

The barrier to generating convincing local news copy is gone. Operators no longer need to understand the nuances of a community to write a plausible story about it. They only need a prompt and a target, leaving local editors to clean up the mess when the synthetic fallout hits their own backyards.

The post AI fake news network invents the collapse of 47 local Alabama newspapers appeared first on The Media Copilot.

Google’s AI-generated search summaries are frontlining details about the recent suicide of a public figure at the top of search results, raising concerns that AI products are amplifying sensitive information that traditional news guidelines advise against publicizing. 

This incident highlights a growing challenge for AI search products as they determine not only what information is relevant, but what deserves emphasis. 

Nicola Agius, CEO of the newsbranch Reach, was the first to identify the issue on June 30 after seeing an Instagram post stating that the brother of Caroline Flack, a former presenter for the U.K. reality show Love Island, had died. Agius searched for “Caroline Flack brother” on Google to verify the news.

Not only did Google’s AI Overview confirm his death, Agius shared that it prominently stated that he had died by suicide, also including the location, method and the effect on his body. The summary also detailed how he was found. 

The Media Copilot was able to replicate the result on July 3. Using the same search query, Google’s AI Overview stated the method of death alongside all facts detailed above. The search was conducted by a reporter in Europe. Results from searches in the United States, however, did not always show the details of his death. 

The Media Copilot also found that even when Google’s AI Overview does not initially surface details of a suicide, simple follow-up prompts can elicit extensive information, including the method, location, police reports, descriptions of the scene and other sensitive information. 

The content displayed by Google’s AI Overview originated in publisher reporting, but they were given greater prominence in the AI-generated summary than in the source articles themselves. In some cases, the information was even removed from the original reporting altogether. One publication cited by Google for the cause of death no longer included those details in the linked article by July 3, while other outlets continued to reference them but with less prominence. 

The Associated Press guidelines on reporting suicide urge publishers “to not go into detail on the methods used.”

“It is incredibly worrying—and disappointing—to see some press sovereign and AI overviews including references to methods of suicide,” said Lois Sparks, head of safeguarding at Mind, a mental health charity, in an interview with Press Gazette. “Not only can these AI summaries be incredibly triggering for people to see so prominently, but in some instances, there is little signposting to mental health support.”

The Media Copilot’s inquiry did trigger a support message, which stated: “If you or someone you know is struggling, support is available in the U.K. by calling the Samaritans at 116 123 or texting SHOUT to 85258. In the U.S., you can call or text the Suicide & Crisis Lifeline.”

“Broadly, AI overviews give an illusion of definitiveness,” Sparks said. “They make highly sensitive and nuanced areas seem like concrete facts. If AI overviews are covering news, then they must be regulated to the same standards of traditional media—which they are currently not.”

The post Google’s AI Overviews surface suicide details, raising questions over AI editorial standards   appeared first on The Media Copilot.

For the second time this year, Google has removed and then reinstated a Press Gazette investigation into Clickout Media from its search results after an anonymous complaint under the U.S. Digital Millennium Copyright Act, restoring it only after the outlet pressed for comment. The delisted article reported that theU.K.-based marketing company had published AI-generated news stories containing factual errors and fabricated information.

The Press Gazette piece, published last week under the headline “AI reporters churn out error-strewn stories for football websites,” reported how Clickout Media acquired three established U.K. football news websites and began publishing stories under AI-generated reporter bylines that Press Gazette found contained numerous errors.

According to records in the Lumen transparency database, which publishes DMCA takedown notices Google receives, an entity identifying itself as “DRF Corp” accused Press Gazette of “willfully” copying its content and images. The complaint claimed the original work was a now-deleted Reddit post. Press Gazette said the allegedly infringing content was unrelated to its investigation. 

The latest takedown follows a similar incident in March, when Google removed a Press Gazette investigation into Clickout Media from its search results after another anonymous complaint. The article reported that the company had acquired news websites to drive traffic to its promotion of online casino content. Google reinstated the story after Press Gazette sought comment. 

The second article has since been reinstated as well after Press Gazette pressed Google for comment. But the pattern of the same target, the same anonymous complaint and a reversal by Google when challenged has drawn criticism from media industry figures, who say bad actors can exploit copyright takedown systems to remove legitimate reporting from search results while low-quality AI-generated content remains visible. 

Dominic Young, chief executive of the micropayment firm Axate and a co-founder of the SPUR Coalition on AI licensing standards, condemned the takedowns in comments posted on LinkedIn. 

“By effectively rendering copyright infringement consequence-free, and reserving the right for tech platforms to profit from it, this law created anarchy online and made copyright infringement into a business model – now being exploited by AI companies and a swarm of proxies helping them get whatever they want, regardless of what the owners say,” Young said. 

The DMCA allows anyone to file a takedown notice regardless of whether they have registered their work with the U.S. Copyright Office. Google reviews each notice to ensure it meets legal and policy requirements. It is not required to remove the reported material, but failing to act on a valid notice could expose the company to secondary liability for copyright infringement, so it usually complies. 

The post Google delists then reinstates Press Gazette investigation into AI-generated news stories   appeared first on The Media Copilot.

Cloudflare said Wednesday it will begin blocking AI training and agent crawlers by default on ad-supported websites, a change that could force companies such as Google, Apple and Microsoft to more clearly separate search indexing from AI training if they want continued access to large parts of the web.

The policy, scheduled to take effect Sept. 15, applies to new Cloudflare customers, new sites added by existing customers and existing Free-tier customers who have not changed their settings. Search crawlers will remain allowed by default, but training and agent crawlers will be blocked on pages that display ads.

The company said the changes are designed to help publishers remain visible in AI-powered search results while preventing their content from being used for AI training or autonomous agents without permission or compensation. 

“Now that the majority of traffic is non-human, we must go further and act faster so that a sustainable ecosystem can emerge,” said Matthew Prince, Cloudflare’s co-founder and CEO. 

Splitting up mixed use crawlers

The Web giant said bots that combine search, AI training and agent activity—known as mixed use crawlers—without letting site owners choose among those uses will be blocked on ad-supported pages when training or agent access is blocked. In a company blog post, Cloudflare named Googlebot, Applebot and BingBot as multi-purpose crawlers that could be affected by the most restrictive applicable rules.

“We hope that our proposed default changes encourage mixed use crawlers to separate out search from agent use and training,” Prince said. 

Cloudflare said customers will be able to manage three categories of AI traffic: Search, which indexes content for later retrieval; Agent, which accesses a site on behalf of a user in real time; and Training, which collects content to train or fine-tune models. The controls are available to all Cloudflare customers, including those on the Free tier.

That distinction matters for smaller sites. A spokesperson for  Cloudflare said the new controls are intended to give all website owners more options for managing AI traffic, not only publishers with ads or subscriptions. But the default blocking policy is tied to pages with advertising, and Cloudflare’s compensation plans remain focused on commercial use cases where AI systems access or surface publisher content.

Alongside the new crawler controls, Cloudflare is expanding analytics to show publishers how bots interact with their content and how much traffic AI platforms send back. The company is also pushing into what it calls Answer Engine Optimization, or AEO, offering tools it says will help customers understand how often their content is cited or surfaced in AI-generated answers.

Cloudflare also announced efforts to reduce unnecessary AI crawling. According to the company, more than half of AI crawler traffic is spent repeatedly checking web pages that have not changed. Because Cloudflare sits between websites and online traffic, it says it can signal to AI companies when pages have been updated and worth revisiting. The company said it is testing those signals with AI firms and plans a broader rollout later this year. 

New compensation model

The company is also expanding its publisher compensation strategy by evolving its Pay Per Crawl program into a new system called Pay Per Use. Rather than paying publishers when content is crawled, the new model is designed to compensate them when their content is actually used in AI products. Cloudflare said it is working with AI companies including Ceramic.ai and You.com on the initiative. Under the arrangements, publishers could be paid when their content appears in AI search results or when AI agents access premium content on demand. 

But the model does not yet answer the hardest compensation question: what happens when a publisher’s work is used for model training but never appears in a cited answer? Asked whether Pay Per Use compensates publishers in that scenario, The spokesperson said the program is aimed at “programmatic, real-time access and discovery,” and described Pay Per Crawl and Pay Per Use as only two possible economic frameworks.

“The digital landscape is evolving rapidly,” said Marrissa Holloway for Cloudflare. “We welcome ideas from publishers, creators, and AI companies alike on how to build a thriving agentic Internet.”

Holloway did not directly say what Cloudflare’s cut of any revenue generated would be. “It has always been our philosophy that our customers derive many multiples of value more than they pay us,” she said.

The Media Copilot’s take

Cloudflare is not solving AI compensation for the whole Web. It’s building a bargaining layer for larger publishers with enough traffic and revenue to measure, block and negotiate. That helps the larger content outlets, but smaller sites and independent publishers will get switches to turn on and off. That’s useful, but switches don’t mean they have leverage. The long tail of the Web—the indy blog sites, community web pages and hobby sites—can say “no” more clearly, but there still no obvious way for them to get paid when their work is used for an AI’s training data and never comes back with a citation or link.

The post Cloudflare will block AI training crawlers by default on ad-supported sites appeared first on The Media Copilot.