August 2025: Spam Update Hits Hard, Global AI Mode & Agentic Google Search

August 2025 Spam Update Hits Hard

Just when you thought it was safe to go back in the water, Google rolled out the August 2025 Spam Update, causing significant fluctuations in rankings within 24 hours of the announcement for websites with outdated or poorly-managed on-page SEO. It's still early days, and the rollout is expected to take a few weeks to complete, but this is already shaping up to be a significant update, hot on the heels of the June 2025 Core Update which was also a big one.

My Take: Spam Updates are regular, routine improvements to Google's spam detection capabilities. One important thing to note (because this is a common misconception) is that these 'Spam Updates' are NOT about spammy backlinks. Google rolls out separate, specific 'Link Spam Updates' when it just wants to target that. These more generic 'Spam Updates' are looking for all types of spam that websites might be using to manipulate search results. If you check out Google's guidelines, you'll get a sense for just how much of what they look for is not link-related... links are just one of about 20 specified types of spam, and they are listed pretty low down the page! If you've been hit, by all means check your backlink profile health, but unless you've ben proactively building spammy links, your time would be better used conducting a thorough audit of your own website for any practices that might be considered manipulative, from keyword stuffing and hidden text to sneaky redirects. In fact, you should let this update serve as a reminder to do that anyway – don't get complacent! The best defence is a good offence, which in the world of SEO means consistently creating high-quality, user-focused content and always reviewing, refining, improving, and keeping up with best practices and emerging UX patterns.

AI Mode Launches in 180+ Countries & Agentic Ecommerce is Coming Soon

Having previously only been available in the US, UK, and India, Google has now expanded its AI Mode rollout to over 180 countries. This is a pretty clear signal of Google's confidence in its generative AI-powered search experience. However, the bigger story here isn't just about answering complex queries with conversational AI; it's about the dawn of Agentic Search – particularly Agentic Ecommerce, which is now looming over the horizon.

With the expansion of AI Mode, Google is testing more "agentic" capabilities, allowing the AI to perform tasks on behalf of the user, such as finding and booking restaurant reservations. As Google's John Mueller has recently (and conspicuously very specifically) advised, it's time for ecommerce sites to start testing their readiness for agentic AI.

My Take: Google is already agentic, if only for a few users and for very limited actions right now. John Muller's advice to start testing your ecommerce sites for agentic AI, combined with the fact that we've recently spotted a "Help me shop in AI Mode" button being tested in traditional SERPs, should not be ignored. This is not futurism or pie-in-the-sky tech talk - it's basically here, right now, ready to launch. If you're a Perplexity Pro user (and we recommend you should be), then you can already download their Comet Browser and start beta testing how your site responds to agentic search bots. We've been testing agentic product search and add-to-basket functionality for our ecommerce clients' sites using Comet, and the results have been incredible—good enough and fast enough that it's hard to imagine users not absolutely loving this if (when) it becomes commonplace in Chrome and/or Google  (and/or ChatGPT). Factor in Google's HUGE advantage gleaned from having its Shopping Graph, which is the largest database of organised, structured product feed data in the entire world, and the potential for this next-gen ecommerce UX paradigm is mind-boggling. 

The AI CTR Debate Gets Even Weirder (or maybe not?)

The great debate over whether AI Overviews and AI Mode are killing click-through rates to websites has taken another interesting turn. After months of the SEO community raising concerns about traffic loss due to in-SERP answers, and Google vehemently denying any impact, Google is now actively testing changes to AI Mode that are specifically designed to encourage more clicks. There's no problem to see here, but we're fixing the problem.

These tests include link carousels and more intelligently placed inline links within the AI-generated answers. This sudden focus on driving traffic back to websites seems to coincide with another impending development: the introduction of ads in AI Mode…

My Take: Of course Google never wanted to admit that they were stealing content and presenting it to users as their own, thereby reducing CTR. And of course Google wants to monetise AI Overviews and AI Mode. And now, as they prepare to launch ads in AI Mode, of course they are doing everything they can to suddenly refine the science of getting people to click those ads so they can maximise their revenue. One thing is for sure… once ads roll out in AI Mode, there will be no hiding the reality around user clicks and conversion, because the PPC model will demand transparency. Silver linings.

It's Confirmed: ChatGPT is Scraping Google Results!

Yeah, you read that right. In a story that falls squarely into the "we knew it all along" category for most SEOs, but sounds completely incredulous to most people outside of the industry, it's now been all but confirmed that ChatGPT uses Google Search as a fallback when its primary sources (ahem, Bing) don't provide the necessary quality of information. This was brought to light through a clever experiment by the wonderful Aleyda Solís, who created a new webpage and observed that ChatGPT was only able to provide a summary of it after it had been indexed by Google, and not before, even when it was indexed by Bing. It was then validated and given credence by the finding that OpenAI is (was?) a customer of a tool called SerpApi, which scrapes Google SERP data!

This revelation, while not shocking to us (based on our own anecdotal evidence) does have interesting implications for the future of search and the interplay between the major AI models.

My Take: If AI responses are 'grounded' by finding up-to-date, relevant info from web searches, are we really surprised that ChatGPT reverts to Google when Bing results don't cut the mustard? No, of course not. This just goes to show that the quality of outputs from even the most advanced AI models are still reliant on the finely-tuned algorithms that Google has built over the last two decades - and that Bing is, well, just not as good. It also raises questions about the long-term viability of Bing as a primary data source for OpenAI, and whether we'll see a future where Google's search index becomes a foundational, and perhaps even a paid, layer for other AI services (anti-trust lawsuits abound, no doubt).

Google Re-affirms Its Stance on AI-Generated Content

One more time for the people in the back… In what seems to be a recurring theme, Google has once again clarified its position on AI-generated content. The message remains consistent: there is no penalty for using AI to create content. The focus, as it always has been, is on quality.

During a recent Q&A, Google's Gary Illyes emphasised that content should be factually accurate, original, and, crucially, reviewed by a human before being published. He suggested that "human-curated" is a more accurate term than "human-created", highlighting the importance of editorial oversight and the value AI can bring when used properly as a tool.

My Take: This is what we've been saying all along. There's nothing inherently wrong with AI content. The specific attribute of AI having been used to aid in the production of a piece of content is not even a consideration. Quality is very much all that matters. The reality is simply that human input is needed to add the required expertise, experience, authority, trustworthiness, and helpfulness, and to anchor and base the content in the best possible way contextually, i.e. so it sits within its website/brand ecosystem with purpose and value. AKA - quality. The bottom line is this: if your content is good, it doesn't matter how it was made. Same as ever.

Google & OpenAI Intent Divergence

A new study from BrightEdge has provided concrete data to back up what many of us have been observing anecdotally: a growing divergence in the types of answers provided by Google AI and ChatGPT for the same queries. The study found that for action-oriented queries, ChatGPT is more likely to provide links to tools and apps, with some analysts noting a significant drop in traffic referrals to sites other than major players like Wikipedia and Reddit – effectively adopting a "task-first" approach. Google's AI, on the other hand, tends to favour an "information-first" approach, linking to articles, guides, and other informational content.

This difference in approach suggests that the two AI giants may be carving out different niches in the evolving AI model landscape.

My Take: I actually predicted this in my last article. These data studies, in my opinion, are likely spotlighting the early phases of a new reality where Google AI is providing informational content as a search surface, while ChatGPT is becoming more of an ingredient brand.​ This is a trend we'll be watching closely.

Bing Launches Copilot Mode to 'Search While Browsing'

True to historic form, Bing has followed in Google's footsteps by launching "Copilot Mode" in its Edge browser – effectively mirroring Google's recent AI-powered Chrome update, which launched in May this year. This new feature allows users to ask questions, summarise content, and perform actions across multiple tabs whilst browsing the web. According to Microsoft, the goal is to create a more seamless and collaborative browsing experience, where the AI acts as a "copilot" to help users make sense of information and complete tasks more efficiently.

My Take: This is another clear sign that the future of browsing is agentic. While this feature isn't new—Chrome has had it for months and there have been extensions available for years that do this—Bing following suit and making native AI browsing assistance the norm is a stark reminder to optimise the search journey beyond the traditional goal-line of "click to site". Search Experience Optimisation (SXO), at the intersection of SEO and UX, is no longer the cool cousin of SEO... it's the whole game.

You need to predict and preempt follow-up questions and next steps. If you don't make it obvious and easy enough, users will simply bounce into the AI copilot to get an easier answer or route forward, potentially taking them to a competitor site or back to search. This forces a long-overdue truth into the spotlight: UX and engagement have always been core components of SEO, just ones that were too often ignored or marginalised. With user tolerance for friction approaching zero, poor engagement being a negative ranking factor, and both search engines and browsers now actively working against you by eroding CTR and session engagement, it's no longer an option to downplay the importance of the UX element of your SEO strategy. The winners are those with the best content, authority, and engagement. That's always been the formula, but hasn't always been the focus of pragmatic SEOs.