September 2025 Google Search Observations

[RubicCubed]

today I have received over 200000 HTTP requests from ChatGPT.

We've been getting hammered with these for over a week. Making matters worse is ChatGPT is accessing URLs they scraped from Google which are appended with different ?srsltid= IDs. We had ChatGPT just hit one of our product pages three times all with a different ?srsltid= appended to the end of the URL. Does ChatGPT actually believe these are different pages? If ChatGPT is going to scrape Google's results, they should at least be smart enough to drop the Google specific ?srsltid= before they visit the URL. Or learn from and respect the canonical versions assigned by us.

[Conro]

Anthropic He will have to pay $1.5 billion to the authors who sued him with the class action lawsuit
[nytimes.com...]

[Martin Ice Web]

BOTS!
since yesterday evening. 1000s of bots.
80% of IP Ranges are located in germany but owned by US IT companies.
20% comes from Israel and India.

Worst ever traffic google is sending.

[seokees]

@Chris travel 30 I target Dutch and Belgium traffic. Last christmas (not the song), between Christmas and New Year I had a similar 'drought'. Good traffic, but 0 conversions. Just at the end of the last algo update from G in 2024. Like, indeed the intent doesn't match with the page visitors land on.

[mosxu]

It looks like AI insight is brainwashing visitors when it suggests other brands when they search for your one specifically like some other brand has 50% off but from what price?

So bad! AI is getting good at moving the cheese now…

[kireb]

I've lost 90% of my traffic to sites that are infringing on my content. I have sent many emails to Google, which they have all rejected. So much for years of good service. Websites that have no contact details, sites that infringe, are rewarded, and I got hammered.

[ruip]

For Google, seems "Content is King" has changed to "Tech is King."
Good content rank hight is now short content rank high.

And the new norm is, "fool me, I like it"... I(google) like large images and structured data more than the originality of the content and the author.

I have websites with 20 years, some pages, like gatormark wrote, is in 3 or 4 SERP results(yes i'm shure is urls with more relevante content for category). Copies of the content is in better place in some results, including original images.

Users from others search engine increase 15-20%.
Seems a end in google search era.
Google is broken.
Can be?

Seems Altavista reading only metatags ... it's seems so 2000

[ichthyous]

Anthropic He will have to pay $1.5 billion to the authors who sued him with the class action lawsuit
[nytimes.com...]

@Conro Don't get too excited. The claimants actually lost the most important part of the case. The judge found that that the "Fair Use" section of US copyright law pertained to Anthropic's use of the publicly available material for training their AI. This means that the authors will not need to be compensated in any way by Anthropic, or any other future AI platform most likely. This is terrible news for all creators of copyrighted material, as it seems that judges are going with the Fair Use argument since the AI is "transformative". Under the law, if the use is transformative enough (and not simply copying) then it is permitted without compensation.

The $1.5 billion was damages for infringement for a huge cache of pirated books that Anthropic had accumulated. It's a huge settlement for sure, but it doesn't help content creators in any way in the fight against AI platforms.

[mosxu]

300 billion deal Open AI and Oracle!

Jump ship now!

[christianz]

300 billion deal Open AI and Oracle!

How is a company that is making less than 0 dollars able to afford a "300 billion deal"? Besides, stealing content is not a viable long term business model. None of these AI answer engines have viable business model, they haven't figure out how to compensate publishers, and when they do, the numbers may not add up.

[RubicCubed]

stealing content is not a viable long term business model.

It is in the USA because the Government has sanctioned IP theft as necessary to win the AI race. In fact the US Government is awarding huge contracts to these companies worth hundreds of millions of dollars even after they are found guilty of antitrust crimes in federal court and continue stealing content.

I suppose once publishers quit publishing and they don't have enough content to steal, these companies will just make up their own content and call it hallucinations. When called out on their made up content they will remind people of their "AI is experimental" disclaimer.

[Fluff_Nutz]

AI can only be experimental for so long. Once the hype has ended people will expect it to work properly or get even more suspicious over it. Also, the US Government only allows copyright theft due to the ''duck'' currently in charge. He is a businessman and not a good president. How he managed to get into that position after so many accusations and criminal activity that was spoken about before he was chosen is beyond me. Guess Americans like idiots.

However, one would assume that once he leaves, which will happen, that someone more capable will take his place and do a much better job. You cannot have copyright laws in place for many MANY years, just to, now, remove them due to personal gain and interest.

[RubicCubed]

someone more capable will take his place and do a much better job

The next President will likely see how easy it was for his/her predecessor to get away with ignoring the law and ignore it even more themselves. Some call it gradualism or the steady state of decay. Politicians are in the same class as CEOs being power and money hungry, so I think it will only worsen and the decay accelerate. We see the steady state of decay in Google as well - Google adds a people also ask box, then a people also search for box, etc. while making results worse so people have to search endlessly within Google to find what they're looking for.

[RedBar]

Big bot hit from Lithuania this morning for my global site otherwise mostly seemingly normal ... Whatever that is.

[saladtosser]

>>>You cannot have copyright laws in place for many MANY years, just to, now, remove them due to personal gain and interest.<<<

Unless of course your a corporate tech entity with investors, then their stance seems to be that laws are for thee, not for me!

[BigKat]

Unless of course your a corporate tech entity with investors, then their stance seems to be that laws are for thee, not for me!

This story [digitalmusicnews.com...] provides a glimpse into how Big Tech CEOs think. These same CEOs are the ones donating millions to politicians in all parties to win their loyalty, and it seems to be working out well for them.

[ichthyous]

However, one would assume that once he leaves, which will happen, that someone more capable will take his place and do a much better job. You cannot have copyright laws in place for many MANY years, just to, now, remove them due to personal gain and interest.

Oh no? We have had an entire government, with countless rules, regulations, oversight committees, etc in place since the end of WWII, and it is all being swept away in one big, well orchestrated coup in nine months basically. The billionaires that own these platforms see themselves as far above the masses. They do not need democracy, it doesn't serve their purposes. They stand to make trillions from stealing copyright.

Once the initial cases are heard and the courts side for the AI platforms then it will make it much harder for anyone else to sue due to precedent. It was always the case that some judges might find AI use transformative enough to shield these companies via Fair Use. But it is open to the interpretation of each judge, and each particular case too. Perhaps the texts used to train the AI were 'transformed' enough to satisfy this judge, but the next judge might not think so. The problem is we give these judges far too much credit as to how astute they are. They hear all kinds of cases, and few are experts on AI or even copyright law. I suspect many will do the easiest thing, which is call it Fair Use and allow the theft to continue.

[not2easy]

Mods' reminder - This forum is where we discuss Google Search and SEO. We do not discuss politics in any of our forums. No one comes here for the latest in politics.

Please stay on topic, it is getting a little close to needing moderation.

[RedBar]

Rightly or wrongly I'm blocking MS Phoenix across all my sites from today, one .com site specifically is visited far too often to be necessary for SERPs purposes. If it were a news site, fine, but no it's an evergreen informational site updated quarterly with amendments where needed.

[christianz]

I think the time when we gave search engine bots red carpet and white glove treatment is over. I have set up hard request frequency caps for all hosts. If they misbehave I permanently ban them on network level. Doesn't matter what type of bot it is.

[RedBar]

@christianz ... I've been considering similar, at the moment I'm looking at Google Dallas as a major annoyance on one specific site, in fact the entire USA is becoming a total pita with bots and so-called spiders!

[mosxu]

@christianz

GPT has the best model, Claude is also good rest of them are absolute bunnies!

One company will self destruct because finding keywords model is dead.

Copyright is not an issue as soon as next year no AI will read anything you write!

[christianz]

One company will self destruct because finding keywords model is dead.

Something is dead for sure because whenever try to search something in supposed search engine (it doesn't have to be Google and I don't have Google as default on any device) using keywords (a string of key words that narrow down the specific topic and what kind of text I want to find), I get irrelevant junk.

Search engines are colossally broken and what they search for is very far removed from what you type in.

Copyright is not an issue as soon as next year no AI will read anything you write!

I hope you are right because I am tired of blocking 12 000 requests from ChatGPT / hour (and growing!).

[courier]

Just noticed, not sure if it is local or in more locations. Up till a few days ago, used to be able to do a search and put &num=100 at the end, your search would list 100 results. Results now seem limited to 10 per page, with less results showing than just the standard search. Suspect it could be to stop SERP tracking tools which maybe require top 100 listings?

search?q=search term&num=100 listed 100 results.

[christianz]

@courier

Of ten you can't even get close to 100 results. It truncates them and shows "We omitted X results that may be not relevant" or something like that. And when you click on "show these results" you still hardly get anything.

In combination with record low search result relevance you often end up not being able to find anything at all.

[Brett_Tabke]

> ?srsltid= IDs.

These are coming from Common Crawl. No one has been able to reason out why Common Crawl is Crawling Google Shopping and getting the srsltid.

> It has been reported ChatGPT is scraping Google's search results to answer some queries

And it has been completely misrepresented.

If ChatGPT has no info:
- in it's own db
- answers in it's common crawl index
- has no other references
- it will use a 3rd party known to scrap Google.

That is what an agentic search browser "should" do on your behalf.

There has been no evidence any where on the web that OpenAi is directly search Google.

I've been following this closely since it was first misreported. I even talked with 2 of the very large news outfits that were prepping stories on it and got them axed, as I showed them exactly where everyone had it wrong.
[searchengineworld.com...]

[Micha]

Conclusion after a few days: Bots are becoming increasingly annoying. My servers are constantly reporting overload because there is another flood of them, no matter how much I block them. Ranking tools no longer work, although the ranking fluctuates extremely anyway, and Google is getting on my nerves more and more.

I really hope that a new provider will enter the market soon and give Google a good kick in the butt (unlikely, I know).

[mosxu]

@Micha

Block the hell out of these bots especially out of your country! White list google classes and bing, gpt if there are …

[tangor]

White list google classes and bing, gpt if there are …

My white list is getting smaller everyday.

As some point (not soon) the web in general will swat the bots for burning bandwidth (which is EXPENSIVE). In short, there will become a BOT WAR---on which BOTS survive and which will die. At the current rate of consumption burn it won't take long before available resources (powered by ELECTRICITY) will be overwhelmed.

</observation, not tin-foil>

[Martin Ice Web]

i am giving up on bots. Like Micha said bots are getting overhand. China, Israel, Singapore, Vietnam and US.
Blocking one seems to trigger 3 new bots.

[mosxu]

If you don’t run product ads countries can be blocked otherwise merchant centre will suspend some products!