Why did Anthropic discontinue its Claude-'penned' blog?

In May, a webpage at Anthropic.com promised its Claude Large Language Model (LLM) was “writing on every topic under the sun.”

While engineering blogs from big tech firms are nothing new, this blog may have been the first that was to be entirely (or mostly) written by an LLM?

But not anymore. The “Claude Explains” blog, from AI research company Anthropic has suddenly vanished mysteriously — and Anthropic isn’t saying why. Was it worried about their legal rights to Claude’s training data? Was the blog making Claude appear just a little bit too human?

Or was it that the blog’s human editors were making Claude look a little less than AI…

Whatever the reason, Claude is clearly being swept into larger debates raging now about potential impacts from LLMs — and also of their potential dangers.

Or maybe this whole improbable episode is just giving us a chance to confront these issues from an entirely new perspective…

Billion Dollar Questions

March saw Anthropic raise another $3.5 billion in investment capital, bringing the scrappy company, founded by former members of OpenAI in 2021, a whopping $61.5 billion post-money valuation, according to CNBC.

Anthropic’s Claude is now already considered one of the most promising LLMs — and by the end of May, it had experienced an unprecedented sudden leap in annualized revenue projections, reports Reuters, from $1 billion to $3 billion. It was that same week that it launched its “Claude Explains” blog. But then something else happened the very next week.

Anthropic was sued by Reddit.

On June 4th, Reddit filed a legal complaint in San Francisco’s Superior Court alleging Anthropic  unjust enrichment, tortious interference, and unfair competition. According to Reddit’s complaint, Anthropic…

• “Trained on the personal data of Reddit users without ever requesting their consent.”
• Ignored the AI-blocking robots.txt files on numerous websites.
• After claiming they’d stopped crawling Reddit, “Anthopic’s bots continued to hit Reddit’s servers over one hundred thousand times.”
• Refused to delete Reddit posts from Anthropic’s systems.

Reddit does license its vast collection of human-written comments to AI companies as training data — but always for a hefty price. Google reportedly is paying $60 million a year, and OpenAI has licensed the data, too.

Rough math, but interesting :) Reddit told AdWeek that its AI licensing deals make up about 10% of its revenue. Annual revenue for Reddit in 2024 was $1.3B, so 10% is $130M. Google pays Reddit $60M so that means OpenAI is paying Reddit about $70M per year. That number has never been revealed. Again, interesting.

— Glenn Gabe (@glenngabe) February 13, 2025

In their complaint Reddit describes its site as “the home of authentic conversation on the internet,” adding that with nearly 20 years worth of comments Reddit “is thus the steward of one of the largest datasets of natural human language discussion in existence.”

And it doesn’t take kindly to being scraped…at least, not without some compensation.

In eloquent legal-speak, the company execs write passionately that Reddit “has never allowed its platform and the countless communities who find a home on it to be appropriated by commercial actors seeking to create billion-dollar enterprises and offering nothing in return to Reddit and its users.” (But with something in return — well, that’s a different story…)

Careful Curation

Then Reddit points to a 2021 paper by Anthropic researchers — which indeed discussed pre-training on large public datasets “such as Stack Exchange, Reddit, and Wikipedia edits”.

Reddit drives home its point with a dramatic courtroom-style flourish. “Anthropic continues to publicly admit that it trains its Al technologies on Reddit content,” company execs write. “And, were there any doubt, Claude confirms as much.”

The complaint then includes a screenshot of Claude’s answering yes to the question “Were you trained, at least in part on Reddit data?”

So now Reddit is seeking an injunction, restitution, and “compensatory damages” as well as “punitive damages” — plus “lost profits and/or disgorgement of Anthropic’s profits”. (And also: attorneys’ fees, and “any other relief the Court deems appropriate.”)

It’s demanding a jury trial.

“We disagree with Reddit’s claims and will defend ourselves vigorously,” an Anthropic spokesperson told The Verge.

But this may already be a sensitive subject within the company. Yahoo Finance points out Anthropic is also facing suits from several music publishers “alleging that Anthropic infringed on copyrights for Beyoncé, the Rolling Stones, and other artists as it trained Claude on lyrics to more than 500 songs.”

And within three days of the lawsuit, Claude’s blog was gone.

Although maybe that’s just a coincidence…

What’s In the Posts?

There’s another possible explanation. You can still read the “Claude Explains” blog posts on Archive.org, where it quickly becomes evident that they’re all oddly technical — and indisputably boring. They’re how-to guides, with titles like “Write reliable unit tests quickly with Claude.”

It’s virtually impossible to mistake these blog posts for anything but what they are: ads for Claude. “How to create a dictionary in Python” includes a section titled “Get unstuck faster with Claude”. (Which clearly has nothing to do with Python dictionaries…) And that post’s final section is titled “Learning or leveling up? Use Claude…”

So it seems that, like any blogger, Claude wrote mostly about itself.

Of course, it wasn’t really writing, but generating text — and most of the posts were ultimately credited to “the Anthropic team.”(So, multiple authors, and not just one hard-working AI…)

Yet they were strangely well-written.

“Performance issues are rarely polite. They manifest as unexplained slowdowns, sudden error spikes, or vague bug reports, and they always arrive at the worst possible moment.

“Claude brings structure, clarity, and speed to troubleshooting…”

And eventually, the human-level quality of Claude’s blog posts raised some concerns.

It turns out the blog wasn’t being written solely by Claude, as TechCrunch discovered. “According to a spokesperson, the blog is overseen by Anthropic’s ‘subject matter experts and editorial teams,’ who ‘enhance’ Claude’s drafts with ‘insights, practical examples, and […] contextual knowledge.'”

So, though the title was “Claude Explains,” the blog was actually demonstrating instead “a collaborative approach,” showing “how human expertise and AI capabilities can work together.”

The first round of tech tips was just the beginning. “We plan to cover topics ranging from creative writing to data analysis to business strategy,” the spokesperson told TechCrunch. But maybe Anthropic changed their mind when their CEO drew headlines with a prediction that AI would eliminate half of all entry-level white collar jobs within five years. While Claude might be capable of generating text for professional use cases, wouldn’t a demonstration now just remind people that AI could displace millions of jobs?

It was just a few days later that an Anthropic spokesperson rushed to reassure TechCrunch that despite its experiment with AI-generated content, Anthropic was still hiring humans, for everything from marketing to editorial positions and “many other fields that involve writing.”

Still, when Claude’s blog was discontinued just days after its launch, TechCrunch wondered aloud if Anthropic had “grown wary of implying Claude performs better at writing tasks than is actually the case.”

Other Possibilities

Or was it really Reddit’s lawsuit that made Claude’s blog suddenly disappear? The circumstantial “evidence”…

• Anthropic is saying nothing — the traditional response when an issue is about to be litigated in court.
• Reddit’s legal documents were filed Wednesday, June 4 — and by Saturday morning (June 7), the blog was gone.

And there’s one final (still circumstantial) clue. Anthropic’s 2021 paper not only included their “whitelist” for high-quality Reddit discussions. It specifically stated that one of them was Reddit’s /programming subreddit — where Python tips and tricks have obviously been discussed.

Was there a danger that all this Claude-generated content might resemble something from Reddit? Even accidentally or coincidentally, that revelation could be disastrous, both in court and in public opinion.

Claude’s blog could be gone for other reasons. (It doesn’t seem it was widely read, or even particularly well-received.) Maybe it was just a marketing idea that ran its course and then disappeared.

Or maybe the designers felt guilty about giving a fake human personality to the output of an algorithm. Some critics have argued that “anthropomorphizing” leads to the over-trusting of systems. A recent article in The Atlantic strongly decried Silicon Valley’s “tradition of anthropomorphizing” (as described in a new book by Karen Hao, the senior AI editor for MIT Technology Review.)

The Atlantic clarifies that LLMs are nothing but “impressive probability gadgets” making “statistically informed guesses about which lexical item is likely to follow another” — but that in the end an LLM simply “regurgitates knowledge without having humanlike intelligence”. Yet when reading text our first instinct is to imagine a normal human mind behind it — an impression some tech executives seem all too happy to encourage. (Linguist Emily M. Bender even titled her book The AI Con: How to Fight Big Tech’s Hype and Create the Future We Want.)

Based on this tendency to ascribe human-like attributes to machine-generated output, the Atlantic’s argues that “the foundation of the AI industry is a scam.”

The New Stack specifically asked Anthropic if they were worried that giving Claude its own blog was “a step too far in anthropomorphization” — and received no response. And on X.com, Claude’s product manager also declined to offer any clues about the blog’s discontinuation.

But it seems like Anthropic may inadvertently get swept up in this debate, since its one of the few leading AI companies that’s given their LLM a human name. Even the company’s own name is Anthropic.

But what’s the real reason Anthropic discontinued the blog? At this point, there’s only one thing that we know for sure. That there’s nobody at Anthropic who can generate an answer for us.

Not even Claude.