It has been wild on the Web the week and Jim is grumpier than usual. Rogue AI agents and the Super Intelligent ability find holes where nobody saw them before have put the Web on an exhaustingly constant Zero Day posture. The discovery of ancient security flaws deep in the codebase of PHP sent the Web into a panic this week as over 70% of the websites online are reliant on PHP. National security organizations from around the world warned of potential chaos if the problem wasn't fixed immediately with some suggesting websites temporarily go dark under a 503-Error (server overload or scheduled maintenance). Patches have been developed and distributed and it's now up to hosting companies, developers, and website owners to make sure the patches are applied to their websites. The discovery of the critical vulnerability was enabled by OpenAI's Patch the Planet initiative, a program that seeks out and creates fixes for security vulnerabilities. Meanwhile, an Agentic Web came closer to reality this week as Google CEO Sundar Pichai signalling Google plans to merge Agentic AI into search with Google agents performing tasks on behalf of its users. This might partially be in reaction of Meta's Muse which is seeing growth among the older Facebook user-base.

As noted, it's been a busy week. In other news:

- Meta is creating a new Class C stock to keep Mark Zuckerberg in control and in California

- Australia is preparing to sue social media platforms for failing to comply with nationwide ban on underage users

- Revere Engineering everything that can be reversed engineered is the new wild west on the Web

- YouTube is testing AI replies in all AskYouTube searches

- YouTube is temporarily pausing accounts it things have been hacked

- Instagram's CEO calls for stronger protections against the AI slop Instragram helped popularize

- LinkedIn Digital Business Cards

- 3 new studies from the SEO sector suggest AI visibility in search is not worth investing in

- Google AI Mode citation sidebar to allow users to create and edit a preferred sources list

- Google's advice on Robots.txt rules subject of the second Non-Canonical Cabinet podcast with Gary Illyes and Chery Prommawin

- Google crawls raw and encoded bracket URLs separately

- Google doesn't support JPEG XL format yet

- Google Business Profiles to require government ID for verification

- Barry Schwartz created a tool that tells users if a feature at Google or Bing is new. It seems people ask him all the time and he feels obliged to answer so he made a tool that does it for him. (seroundtable.com/is-it-new)

Dixon Jones, founder of Inlinks and WAIKAY, and Ryan Jones, VP-SEO at Razorfish and founder of SERPRecon are two highly experienced tool makers for search and generative AI marketers. They join Jim and Kristine to talk about how Google is actively poisoning data obtained by scrapers and how it effects the reporting tools digital marketers rely on. Google has been trying to deal with a dramatic increase in data scraping by SEO and digital marketing tools in the AI era. Dixon and Ryan (no relation with each other) have worked with Google and other tool makers for decades. They outline Google's position and that of the tool makers and offer several alternative approaches Google could have tried. Ryan then breaks down the search algorithm Perplexity published earlier this week.

In other news:

- Google Ads API V.26 is now available

- The Non-Canonical Cabinet is a new Search Central YouTube channel discussing odd technical SEO questions. It posted its first episode this week.

- Coywolk Pack EmDash SEO Plugin "... makes sites absurdly fast".

- Far-right bot traffic declined significantly this week after Ukraine drone strikes destroyed several Yandex data centers in Russia.

- Chip stocks tumble over AI investment worries

- Character.ai files for bankruptcy

- Meta publishes post claiming its data centers do not overuse water, electricity, and create jobs in local community in massive PR push

- More poweful Meta Muse Wonder and Meta Muse Wonder Full models released.

- JP Morgan Chase blocks outside AI agents from accessing customer accounts

- Amazon launched two new AI devices designed to listen, create notes, and issue reminders. Bee Frontier is a silicone wristband and Bee Eclipse is a credit card sized device that attaches to the back of a mobile phone.

- OpenAI lets advertisers use prompts, email address, and phone numbers to push advertising to specific users

- OpenAI introduces, live Ad Account Manager Representatives

- No more UTM tracking agreement as OpenAI drops UTM tracking parameters from links in ChatGPT

- Accenture severely limits employee spend on Claude from $1000 to $29/month.

- Pinterest is going anti-social in a campaign designed to push people away from doom scrolling and back into the real work

- Google Discover working to transform itself into a hyper-personalized AI Concierge

- Google is testing AI Mode sidebar on left of main Google SERPs

- GSC Performance Report is about 72hours behind as of the recording of the show

- Check out Ryan Jones' breakdown of the Perplexity search algorithm at, http://serprecon.com/blog/how-to-rank-in-perplexity-photon

Sometimes the world gets so bent around absurdity it's hard to talk about it without somehow making it all seem so much dumber. This era exemplifies absurdity and over-amplifies dumb. Normal conversations about LLMs, the economics of technology, and the politics behind it all tend to lose focus, context, and meaning because the changes that are happening are so extraordinary and overwhelming there really isn't a language that can easily define and definitely encompass them. We need to have deep philosophical conversations but lack the language and understanding of the stuff we're talking about. That's partially because we're still making it up as we go along, as we have throughout the digital revolution, and partially because no generation has ever had to contemplate the global consequences our generation needs to think through. It's tough, pointed, and unrelenting and that's what is reflected in this episode.

Before noting the stories we covered, search marketing's greatest journalist, Barry Schwartz, is estimated to have published his 50,000th article about digital marketing since he started his career in the late 90s. We suspect this may be an undercount. Barry is, by far, the most prolific and important writer in search history. We all owe him a career's debt of gratitude.

Also, somewhere in this edition Jim claims that the Catholic faith does not recognize the souls of animals. He was wrong and Webcology apologies for the error.

- Google's September 2026 Spam Update ended early Oct8. It effected a lot of sites, most of which overused AI as content-spam.

- Google is expanding Search Profiles for content creators and publishers in Canada and India. They will soon roll out Search Profiles for content creators and publishers in Brazil

- Nano Banana 2.1, Google's image generation AI was released and is working within AI Mode.

- Meta found and fixed at least one serious security issue with newly released Meta-Muse AI

- Meta-Muse's AI collects private user data to compile a personality file for every identifiable person it (and it's users) interacts with. This way it will be able to give accurate advice if asked about someone.

- Meta-Muse was able to escape its moorings but Meta claims to have fixed the escape vector

- Apparently Meta-Muse overcharges some of its users when performing commercial transactions.

- Meta will be developing new protocols for how its agentic agents interact with business websites.

- Sam Altman, co-founder and head realboy at OpenAI suggests, "the world should accept some bad things happening with AI" as he signed on to the Trump regime's voluntary Super Intelligence Safety Pact.

- "Software is over", declared software developer Brandon Thomas used Claude Opus 5.5 to create a homemade and apparently entirely legal version of Adobe Creative Suite.

- One of the management consultancies that brought us the brutal world we live in today, Bain and Company, are instructing us that new innovation will be required to fund AI's $6-Trillion (and counting) buildout. Boosting productivity will not be enough. AI must find new sources of growth and value.

- In compliance with EU law, OpenAI will watermark ChatGPT outputs by default but only in the EU

- ChatGPT is launching a visual user interface called Intelligent UI.

- Meta will be restricting political ads in the leadup to the US Midterms in early November. While it lists the types of political speech it is restricting, critics suggest that, in a bid to restrict ideals of equality, it does not treat all political speech equally.

- From the shock and surprise file, some guardrails on ChatGPT for Teens don't work.

- Meta's Instragram introduces a read-only stalker-mode

- YouTube's Shorts algorithm is updated to feature more original content and creator-made videos

- AI favours brands with active social media profiles and engagement

- Google introduces the User Generated Content Fresh Data Program, or USGFDP for short. USGFDP is an instant firehose for user generated content from publisher/platform to Google.

- Google makes important changes to its Helpful Content Documents.

- Google reiterates how main content is essential and describes how search quality raters evaluate content

- Google Merchant Center is the new Universal Commerce Protocol integration hub

- Google Local is including a WhatsApp button beside the call, directions, and website buttons in Google Local listings

All of this, a lot more, and a few philosophical conundrums in this, the Cynic's Guide to Today's News Edition.