Uh, go a page or so, looks OK. But THEN it hits you with a login page…
Nature isn’t a tech publication so I don’t blame it for not delving into the details. I’d say the article is describing what is essentially a denial-of-service attack. In any case, website scraping has gone on since the launch of Google search (as opposed to the earliest Web search sites which relied on human compiled lists) but has reached unprecedented heights with the advent of generative AI. So this means Meta, Google, OpenAI, Anthropic, Apple, and many other companies are “who is behind it”.
Currently, I believe most of the defenses are similar to those used to guard against denial-of-service problems. But I wouldn’t be surprised if any site that depends on advertising revenue and wants to participate in Google, Meta, or Adobe’s ad networks is required to leave its site open to genAI scrapers.
1 Like
Hi Eric,
Just for the record, I am 65 next year, so remember programming BASIC with punched tape on an HP PDP 11.
I saw my first Mac in 1985 and loved it but at £3000 in the UK, it was definitely not a computer for everyone!
I eventually got an LCII in 1992. I have persuaded about 50 people to switch to Apple and provide free tech support to all of them. I wouldn’t say I was any busier with that now than I was, say, 10 years ago.
Apple has many complex issues to deal with and still manages to amaze me occasionally. If you use Macs, constantly running the brand down is ultimately self-defeating, as you may drive users to the alternatives. Apple could easily collapse, despite its apparent wealth, just like Kodak etc.
I have not found anything better, to date.
I’m not familiar with the “HP PDP 11”. Did you mean DEC PDP-11? Or HP-1000?
1 Like
I agree with your general interpretation of the phenomenon functioning essentially as a DDOS attack. Unfortunately, the targets mentioned in the article (public research databases, academic journals, etc.) are particularly ill-suited to combat such attacks.
I’ve had my quibbles with Nature, but to be fair to them, the article was from their “News” section, which tends toward short, high-level summaries. They’ve done a good job covering AI issues in general, though. For example, I think their Dec 2023 feature article asking “Is AI leading to a reproducibility crisis in science?” should be mandatory reading for any scientist using AI/ML, especially life scientists.
As a journalist (who has written for Nature), I see their story as breaking news, which by its nature covers immediate impact rather than does a detailed analysis. I agree that the immediate impact resembles a denial of service attack, but the targets of AI scraping are data archives, journals and publishers which are very important to the scientific community. Nature itself is a major scientific publisher, so this kind of attack is of particular interest for Nature itself.
Once a story like this has broken, you can expect others to follow that go into more detail about the severity of the problem, its effects, who is responsible, and possible countermeasures. I’ll be watching that to see what happens. Your comment on the possible impact on advertising is interesting. Big advertisers tend to carefully track the effectiveness of their ads, so they may may be among the first to react if affects them.
2 Likes
Here’s a newer use of Large Language Models-Generative AI that probably will have a huge effect on academic journals and scientific archives:
While conventional chatbots spit out information in static responses to prompts, AI agents are different. After being queried, they can handle complex tasks on their own, without step-by-step instructions. In theory, they should come much closer to mimicking a real human assistant.
Manus, for example, can hoover up data from the web in real time and then analyze it. It can compile that information into documents, tables, graphs or spreadsheets. And it can provide all the files and code required for an entire website, or even host that site itself at the click of a button.
https://www.bloomberg.com/news/newsletters/2025-06-03/china-s-manus-shows-the-promise-of-ai-agents-review
I betcha AI Agents, if they take off, will make the environment for teachers even worse that it is now…
Apologies, Brad. It was definitely a PDP-11, so it must have been a DEC. It was about 50 years ago!
Not a problem. If you want to revisit history (well, minus the paper tape punch) see: PDP-11 Replica: the PiDP-11
I have one myself, though I’ve mostly been spending time with the PiDP-10 on my infrequent visits to the past 
1 Like
For anybody interested, updates on what is happening at MacInTouch are posted here (as of today, fending off bots not blocked by Cloudflare is the focus):
https://www.macintouch.com/updates.html
2 Likes
macintouch.com is currently down due to major Cloudflare and Kinsta problems:
Ric Ford
(Please note that I’m not actively monitoring and participating in the forum but may rarely post significant news updates related to macintouch.com.)
4 Likes
Looks like this is a much bigger problem, involving Google Cloud Services and therefore a huge chunk of everything on the Internet (including Cloudflare):
3 Likes
Bleeping Computer has an analysis today with quotes from Cloudflare about an unnamed “third-party cloud provider outage.”
The issue has been largely mitigated. It started 17:52 UTC yesterday when the Workers KV (Key-Value) system went completely offline, causing widespread service losses across multiple edge computing and AI services.
Workers KV is a globally distributed, consistent key-value store used by Cloudflare Workers, the company’s serverless computing platform. It is a fundamental piece in many Cloudflare services and a failure can cause cascading issues across many components.
The disruption also impacted other services used by millions, most notably the Google Cloud Platform.
(It’s not clear if this situation was connected in any way to events in the Middle East.)
1 Like
More Cloudflare problems today:
1 Like
2 Likes
So that begs the question, what is an “API management issue”? They failed to throttle calls to some API(s) that far exceeded the conditions of use? The API went rogue due to malformed access calls?
1 Like
Google has explained things here:
https://status.cloud.google.com/incidents/ow5i3PPK96RduMcb1SsW
Since Dia has just become available to Arc users and this seems like exactly what they have in mind for it, I asked its chatbot:
Explain this report to someone who uses technology but is not an expert in highly scalable Internet services.
On June 12, 2025, a major Google Cloud outage caused widespread problems for many online services worldwide.
Here’s a clear explanation of what happened, why it mattered, and what Google is doing about it:
What Happened?
- A software update went wrong: Google made a change to a system that manages how their cloud services check usage limits (quotas) and permissions.
- A hidden bug: The new code had a bug that wasn’t caught during testing because it only appeared when a specific kind of policy change happened.
- Bad data triggered the bug: On June 12, a policy update with missing information was sent out. This caused the system to crash in every region where Google Cloud operates.
- Global impact: Because Google’s systems are tightly connected and changes spread quickly, the problem affected users and businesses all over the world within seconds.
What Was the Impact?
- Many Google services failed: Products like Gmail, Google Drive, Google Cloud Storage, BigQuery, and many more had trouble. People and businesses couldn’t access data, run apps, or use APIs (the tools apps use to talk to Google’s services).
- Error messages: Many users saw “503 errors,” which means the service was unavailable.
- Some regions recovered faster: Most places were back online within a couple of hours, but the central US region took longer because the recovery process overloaded another part of Google’s system.
Why Did This Happen?
- Lack of safeguards: The new code didn’t have enough error handling or “feature flags” (tools that let engineers turn off new features quickly if something goes wrong).
- Testing gaps: The bug wasn’t found in testing because the exact situation that triggered it didn’t happen until the real policy change was made.
- Fast global updates: Google’s system is designed to update everywhere quickly, but this meant the bad data and bug spread instantly.
What Is Google Doing About It?
Google has promised to:
- Add more safety checks and error handling to prevent similar bugs from crashing their systems.
- Use feature flags for all critical changes, so new features can be turned off instantly if needed.
- Improve testing, especially for how new code handles bad data.
- Make sure their communication and monitoring tools stay up, even during outages, so customers know what’s happening.
Why Does This Matter?
If you use any service that relies on Google Cloud (which is a huge part of the internet), you may have noticed problems or downtime on June 12, 2025. This incident shows how a single bug in a widely used system can have a massive ripple effect, affecting millions of people and businesses around the world.
If you want to read the full technical details, you can visit the Google Cloud Service Health page.
5 Likes
Thank you for the analysis of Google’s full incident report. Mea culpa: The link Ric posted above did have more information on it than I thought. I made the classic error of not reading below the fold. To borrow a phrase, “we will [try to] do better.” 
1 Like
James, I was, am, the opposite. Perhaps bias as I once had an article in the physical MacWeek Macintouch magazine.
Foremost, it is Ric’s site and frankly, he was spot on about Apple’s move to subscription, and how Apple stores killed privately-owned mac enthusiast support stores. Like any rebuke, perhaps you create your own Apple support, technology and forum site. Let us know how it goes! And how websites getting clicks for rumors did much better than technical information. You should be critical of a trillion-dollar company that dropped enterprise support, pushed its own support site over 3rd party, that collects data on use and non-use by its customer base (oh the data isn’t resold, but the data decided that you get a model with only 2 USB ports and no headphone jack). People forget, and Apple has made lemons, dropped really good hardware (Time Capsule could have been integrated into Homepod!), pushed subscriptions and Applecare+, and all for what? Greed. Even now, look at all the grey haired directors at Apple. Just look how Apple bought out many of its 3rd party (under NDA so I can’t really say how many) devs. Most recently, what irked me was they bought Affinity! I mean, the best affordable alternative to Adobe apps! Will then next OS have it like Keynote, Numbers, Pages and maybe Imaging?
However, I think we all can agree that many of the headaches of Apple’s previous generations of computers are all but gone. No longer rebuilding databases, repair of fonts, SCSI chain issues, failing optical drives, GPU issues, … yet I still bash Apple for its annoying App store ads when you search for, e.g. Outlook and the first result is not Microsoft Outlook but a clever ad.
Speaking about bye-bye to the Intel imac support… have you ever had to move one, say from a floor to another, carrying it by hand? To unplug all the cables and then pick it up? How do you do that without cutting into the meat of your palm? The edges on that iMac are horrible. Can’t carry it by the foot, as it swivels. Sides? good luck. Nick the side to a wall and crack goes the non-tempered glass. Now, pick up an M-series iMac and its like nothing.
(my work involved deprecating those 27" monsters to be…well…crushed into recycling material).
Note: I understand iOS 26 will FINALLY have a toggle to prevent dialing recent numbers accidentally. IT will require a few more steps like taking you to Contact, and then calling from there. Only took 19 (ergo 26) evolutions of an OS…
1 Like
Apple does not own the Affinity graphics products nor its developer (Serif, acquired by Canva about a year ago).
1 Like