OpenAI has launched its new browser Atlas built to compete with Comet, Chrome, Edge, and other Chromium based browsers. As with other AI based browsers, Atlas comes with a slew of amazing self-directing features along with the potential for a long term mess of major security flaws, exploitable bugs, and the threat of malicious prompting. These are the earliest generations of AI based browsers so both problems and rapid improvements are inevitable. While AI is being added to virtually everything, two federal judges warn it should not be used in law noting how judges and clerks using AI in their writing have led to serious errors in US court rulings. Meanwhile Microsoft has added Harvard Health sourcing to Copilot. Reddit is suing Perplexity and SerpAPI over their scraping of Reddit data from Google's search index, which contributed to Google's decision to severely limit the size of results sets available to APIs. We get more information about impression loss at GSC. Google notes that links, technical SEO, and migrations can't fix craptastic quality issues. We're assuming they're talking about content but not being as clear as possible. Research shows LLMs are used for research and information and websites are used for buying as conversions from LLM traffic tends to be lower than those sent by Google search. All this and more on a truly browserific edition of Webcology.

Google is experimenting with several AI additions to several more Google products as it fixes bugs in its organic Google search product. It is getting to feel like SEOs are getting a better handle on how to work with and affect AI generated responses as generative AI reponses find their ways in between a greater number of user experiences. A small survey by Kevin Indig revealed a lot about user behaviors and AI drawing conclusions which included; AI Mode is sticky, clicks from generative AI responses tend to be transactional, and AI Mode matches site type with search intent. Another survey by enterprise platform Yext shows AI responses tend to rely on information from brand controlled sources rather than Reddit or YouTube. SEO researchers have found several ways to manipulate AI models, one of which is to fake publishing dates for better "freshness". Before getting too fresh with AI, one should consider former Google CEO Eric Schmidt's fear that AI models can be hacked and used for nefarious and even dangerous purposes by bad actors. Regardless of Eric's trip, Google's lawyers and lobbiests are working hard to convince the courts to allow Google to add Gemini to virtually everything by not punishing the company further in the ongoing Anti-Trust settlement.Gemini is close to taking control of Chrome Browsers along with Windows and Andriod operating systems while OpenAI is working on transforming into an operating system itself. Google has added NotebookLM to its list of user-triggered crawlers or fetchers. The WordPress legal saga is back on again, Microsoft talks about writing and preparing for AI crawlers, and Google apears to be again expanding the heavy lifing Google Business Profiles is doing with the introduction of GBP Insights. All this and lots lots more...

Jim frets as his birthday comes up for renewal again next week. Meanwhile, Google's head of search, Elizabeth Reid was featured on an Economic Times podcast last week talking about the future of search, the effects LLMs have had on both search and user behaviors, the expanding roles of Agentic AI, and how important India and its extraordinary wealth of tech talent will be to Google's future. OpenAI announced its new Agentic Commerce Protocol which gives merchants a way to build agentic commerce experiences with ChatGPT users effectively bringing instant checkout to ChatGPT. Bots might be represented as traffic but a TollBit study shows Google sends 831 times the number of actual visitors to webites than AIs do. Video search is high in AI searches with YouTube being, by far, the dominant channel. Perplexity's AI search driven Comet Browser is now free and available to all users. Disney characters are being removed from sometimes sexually charged chatbots just as Tilly Norwood makes an appearance as a AI character created by production studio Particle6. Meta is going to sell targeted ads based on users' AI chats. Google is introducing a Gemini powered Google Home speaker/vocal interface in spring 2026. Google AI mode is getting better at analysing images and has introduced agentic features to US based users. Google Ads Reporting might have been affected by the num=100 change including possible issues with keyword recommendations. All this and far more on a preparty edition of Webcology.