TUAW Joins iLounge as an AI-Powered Zombie Site 

On Mastodon, Christina Warren revealed that The Unofficial Apple Weblog, better known as TUAW, has been reanimated after being defunct for a decade. There’s nothing inherently wrong about buying a domain name and reviving a website, but the new owners, Web Orange Limited, kept the names of the original writers alongside AI-generated photos and bios and atop what seem to be AI-rewritten articles from the TUAW archives. In response to a complaint from Warren and the public outcry, Web Orange Limited changed the names, though in at least one case merely by changing a first name, turning well-known developer Brett Terpstra into zombie Paul Terpstra (no, not that Paul Terpstra).

In fact, the entire TUAW site seems to be powered by AI. Web Orange Limited says on the About Us page that it acquired TUAW “without its original content” but notes:

With a commitment to revitalize its legacy, the new team at Web Orange Limited meticulously rewrote the content from archived versions available on archive.org, ensuring the preservation of TUAW’s rich history while updating it to meet modern standards and relevance.

Christina Warren said AOL shuttered TUAW in 2015 and moved the archives to Engadget, breaking all the URLs. I suspect Web Orange Limited used a spider to crawl all old TUAW URLs on the Wayback Machine, had an AI tool “meticulously rewrite” those articles, and posted them on the new site. That would explain how we get Steve Sande’s favorite apps article from 2014 written by “Jeffrey Adams” but still using the first person. Compare against the original on the Engadget site.

TUAW was prolific, so there are thousands of articles going back to the site’s founding in 2004. When I checked out the oldest pages, I quickly found some places where the system broke down, linking modern headlines with ancient content and revealing what looks like generative AI behind the curtain.

Why do this? There are a lot of old TUAW URLs floating around on the Web. They didn’t resolve a week ago, but they do now, and their content will often pass the average Web surfer’s sniff test. Having content on the other side of all those old URLs will help the overall SEO ranking for the TUAW site, thus increasing the chance Google will return TUAW pages for searches and cause people to see ads.

To that end, Web Orange Limited hasn’t restricted itself to resurrecting obsolete content. With a quick jump from 2015 to 2024, current articles have started to appear on the site. Curious as to their provenance, I fed the headlines of several articles to Google, and in many cases, one of the top hits was a MacRumors article. As you can see below, the TUAW article appears to be based on the MacRumors piece.

So much for that whole “editorial integrity” thing from the TUAW About Us page.

So what is Web Orange Limited, anyway? The company is incorporated in Hong Kong, and its website says that it is an online advertising agency. Although it claims big-name clients such as Avast, Adobe, and ExpressVPN, the site leads with “Our Brands,” listing long-time Apple site iLounge alongside others with which I was unfamiliar. Web Orange Limited says its sites collectively attract over 10 million readers per month. We linked to various iLounge articles between 2005 and 2010, but my sense has been that the site has gone downhill in the past few years.

I wasn’t wrong. In January 2019, iLounge founder Dennis Lloyd sold iLounge to Web Orange Limited. That sale included all the content, so many iLounge URLs continue to resolve to their original articles. Others redirect to unrelated articles dated after the 2019 acquisition.

Given that generative AI became widely accessible starting in 2023, the iLounge articles between 2019 and 2023 must have been created another way. Summarization tools have existed for many years, including the Summarize service in macOS, so iLounge could have used such a tool or hired people to recast text. This 2021 article about the “Lessons in Chemistry” Apple TV+ show looks suspiciously similar to another MacRumors article.

With the exception of zShare, which I couldn’t track down, all the other sites listed as brands on Web Orange Limited’s site are now nothing but random, seemingly AI-generated posts. Of these, Metapress, Tapscape, and The Hack Post were been digital content publishing companies; Soup.io was an Austrian social networking and microblogging site; Ziddu did something with warehousing and blockchain; and I couldn’t dig up any background on IMC Grupo.

What should we take away from this situation?

• Avoid TUAW and iLounge: Both sites would appear to be scams designed to exploit other people’s work to generate ad impressions. Content from before 2019 on iLounge is still real, but everything else at both sites is suspect.
• MacRumors may want to fire up its lawyers: It’s unclear if what Web Orange Limited is doing violates copyright, but given the pattern of behavior and the use of original art alongside rewritten text, at least a warning letter may be worthwhile.
• Generative AI is complicit but not responsible: It’s tempting to blame AI for this, but remember that iLounge has been posting rewritten content since 2019, well before ChatGPT appeared. Generative AI deserves neither all the hype nor opprobrium it receives.
• Google and other search engines are on the hook: This scam works only because TUAW and iLounge had decent reputations, both with the Apple community and in Google’s search rankings. AI slop, if that’s what this is, isn’t necessarily worse than human-generated plagiarism, but it can be generated at an unprecedented scale. It’s up to Google and other search engines to detect such sites and ensure they don’t bubble up to the top of search result pages.
• Stick to trusted sources: Although search engines will always play a role in discovering Web resources and many people will continue to click random links fed to them by social media algorithms designed to increase engagement, I believe the best strategy to avoid AI slop and human-created crud (see Sturgeon’s Law) is to focus on trusted sources. Reject the algorithms and get your information from reputable publications, well-known individuals, and recommendations from either.

In the end, situations like this are just depressing because everyone other than the scammer suffers. Readers receive poor information, publications and authors lose traffic, and search engine reputations are damaged.