provenance

Eric Frydenlund’s recent column arguing that “AI lacks the authenticity of human experience” landed as both a provocation and a practical warning: provocative because it challenges a booming tech narrative that smarter models equal better judgement, and practical because the failures Frydenlund...

A new University of Sydney analysis shows Microsoft Copilot’s AI‑generated news summaries systematically under‑represent Australian local and independent media, favoring large national and international outlets instead — a pattern that risks diverting referral traffic, eroding byline visibility...

Microsoft’s Copilot ecosystem appears to be moving toward built‑in provenance signals for AI outputs—most visibly for images—with multiple signals pointing to watermarking and C2PA content credentials being applied across Microsoft services, while broader claims that Microsoft 365 Copilot will...

The experiment described by ZDNET — asking six popular AI assistants the same set of trick questions and watching every one produce at least one confident-but-false answer — is not a sensational outlier; it’s a precise, reproducible snapshot of a structural weakness in contemporary...

The tidy, confident prose of mainstream AI assistants still hides a messy truth: when pressed with “trick” prompts—false premises, fake-citation tests, ambiguous images, or culturally loaded symbols—today’s top AIs often choose fluency over fidelity, producing answers that range from useful to...

OpenAI’s Sora 2 makes it remarkably easy to turn a short home video into a tiny cinematic — and a recent hands‑on test that used the tool to generate cameo clips of a pet cat shows both how magical and how delicate that power can be. The experiment produced photorealistic, lip‑synced short...

There is no single “best” AI chatbot — and that’s the most important takeaway from a year of head‑to‑head testing that put Claude, ChatGPT, Gemini, Copilot, Perplexity and others through dozens of real‑world tasks judged by humans. The practical winner is not one model but a tool‑for‑task...

John Donovan’s two December 26, 2025 postings on royaldutchshellplc.com — framed as “Shell vs. The Bots” and a satirical “ShellBot Briefing 404” — are not merely another chapter in a decades‑long personal feud; they are a deliberate test case for how adversarial archives interact with modern...

The gaming industry’s simmering debate about generative AI erupted into public view this month when an awards season juggernaut — a critically celebrated indie RPG — triggered a chain of denunciations, a rescinded prize from a niche awards body, and renewed calls for disclosure and limits on...

Microsoft’s engineering gamble — to use AI to rewrite millions of lines of legacy C and C++ into Rust by 2030 — landed squarely in the spotlight this winter after a months‑long string of Windows 11 malfunctions and a formal Microsoft support advisory that traced the outages to XAML registration...

PCMag Australia’s roundup of “The Best AI Image Generators for 2026” captures a practical moment in a fast-moving market: a small set of feature-rich, web-accessible services now dominates everyday creative workflows, but the real story is about trade-offs — integration, moderation, provenance...

Microsoft and NASA’s new Hydrology Copilot promises to make high-resolution hydrology data accessible to non‑specialists, turning dense satellite and model outputs into plain‑language answers, traceable maps, and exportable datasets that can speed flood risk assessment, drought monitoring, and...

Morningstar and PitchBook have quietly but decisively moved a large piece of financial intelligence into the conversational AI layer: their new Model Context Protocol (MCP) apps are now available inside ChatGPT, letting licensed users query analyst-backed public and private market data directly...

Merriam‑Webster’s pick of “slop” as its 2025 Word of the Year crystallizes a cultural turn: after three years of explosive generative‑AI adoption, public vocabulary finally has a single, scornful word for the glut of low‑quality, mass‑produced digital content flooding feeds and search results...

Generative AI’s arrival in art, media and entertainment has been messy, consequential and rapidly pragmatic: a wave that began as an existential threat in 2022 has already re-shaped business models, courtroom strategy, and creative workflows — and today many creators are learning to combine...

Generative AI chatbots are increasingly producing polished but fictional research records — invented journal titles, bogus DOIs, and archival catalogue numbers — a problem the International Committee of the Red Cross has publicly warned about and that recent audits and peer‑reviewed tests...

360WiSE’s December 6 press announcement claiming that multiple AI systems — including Perplexity, Microsoft Copilot, Google AI Overview, OpenTools and ChatGPT — have identified the company as a “rising global media authority” crystallizes a new battleground in modern discoverability: the race to...

On 29 October 2025 John Donovan published what he says is the unredacted transcript of a conversation with Microsoft Copilot about Royal Dutch Shell’s ethics — a public flashpoint that crystallises three decades of commercial disputes, a sprawling online archive of leaked documents, and an...

On 6 December 2025 a piece published on royaldutchshellplc.com set off a small, but revealing, chain reaction: Elon Musk’s conversational model GROK produced a confident biographical sketch of veteran campaigner John Donovan — and in doing so asserted a factual claim about Donovan’s late father...

The latest consumer-facing audits and public‑service studies paint a stark picture: mainstream AI assistants are regularly making repeated factual errors, misattributing sources, and presenting confident but unreliable guidance — problems that matter now that these systems are embedded into...