169 items, newest first. Page 2 of 8.
Anthropic's J-Lens reads Claude's hidden reasoning layer, exposing a gap between observed and actual model behaviour.
35% of web-search LLM answers flagged for untrustworthy citations in University of Iceland pre-launch audit.
Cloudflare's three-tier AI crawler policy, live September 15, changes which LLMs can see your content.
Self-promotional listicles win LLM citations now, but accelerate their own credibility discount.
AI search agents score worse searching ambiguous queries than guessing, per DiscoBench. Content precision is now a brand-visibility lever.
Claude Sonnet 5's token inflation nearly doubles real task costs despite unchanged list prices.
An OpenAI genomics paper accidentally names three GPT-5.6 Pro variants, signalling a new tiered model architecture.
Hallucinated citations are passing peer review at top AI venues, corrupting the corpora that LLMs learn from.
New data shows AI Overviews cut full-value traffic, not just low-quality clicks.
LLM groupthink rewards high-volume publishers. Institutional knowledge in PDFs barely registers.
ChatGPT's real source logic, read from network traffic, not outputs, rewards third-party validation over owned content.
Anthropic's Claude Science brings autonomous research AI to pharma, with real consequences for which brands get cited.
Claude Sonnet 5 matches Opus-class quality at lower cost, accelerating enterprise AI deployment and LLM citation exposure.
Anthropic's Sonnet 5 beats Opus 4.8 on knowledge-work benchmarks, reshaping enterprise model selection.
GPT-5.6 Sol's stronger reasoning will lift citation thresholds, rewarding primary-source authority and penalising aggregated content.
OpenAI's three-tier GPT-5.6 pricing means Luna, not Sol, will handle most enterprise queries about your brand.
Mueller explains AI impression counting in Search Console, revealing a gap between being cited and being seen.
LLMs cite third-party sources for brand information six times more than owned ones, per new arXiv research.
AI constructs different brand reputations depending on query language, a 35,640-response study finds.
New indices measure which brands own LLM category answers, and most competitors don't appear at all.
Google confirms third-party GEO tools lack its internal metrics, leaving brands to optimise against proxies.
A German court found AI Overviews are Google's own speech. B2B brands with sloppy content now carry new risk.
Google's preferred sources feature locks established publishers in and keeps new entrants out of AI Mode answers.
100K prompts confirm a three-tier brand ladder in AI search, with global names appearing in 73% of relevant answers.