In this chapter, Scott and Reece rummage through the digital junk drawer of the last fortnight. From OpenAI’s relentless updates and Gemini’s experimental leaps, to Meta’s surprise power move and one AI-fuelled New Year’s Eve disaster, they stitch together a narrative from chaos, curiosity and very real consequences.

In this episode:

🤖 ChatGPT 5.2 Lands (Again) – GPT-5.2 quietly drops and immediately outperforms 5.1 in real classrooms, showing more personality, better reasoning and stronger challenge — at the cost of speed. Free users also lose auto-routing as OpenAI tightens usage controls. Read More

🖼️ ChatGPT Image Gen 1.5 & Adobe Apps – OpenAI hits its own “Nano Banana moment” with vastly improved image editing, plus a surprise app ecosystem including Photoshop, Acrobat and Express — free previews that funnel power users into paid tools. Read More

🏪 The ChatGPT App Store – Airtable, Notion, Stripe, Canva, HubSpot and more arrive as first-class integrations, letting ChatGPT pull from — and push into — real business systems. Another step towards being the homepage of the internet. Read More

🌐 Gemini Draw-to-Edit & Experimental Browsers – Google adds scribble-based image editing, Disco Browser concepts and Gen Tabs, hinting at a future where AI builds apps directly from your open webpages. Powerful — but still chaotic. Read More

📚 NotebookLM Evolves (and Struggles) – Gemini fully replaces NotebookLM’s old model, adds data tables and structured outputs, but loses some of the laser-focused clarity power users relied on. Progress with friction. Read More

🧠 Super Gems & Task-First AI – Gemini introduces Super Gems using Google’s Opal engine, creating mini AI apps with custom interfaces — pushing closer to Claude’s artefacts and task-oriented workflows. Read More

🧑‍💻 Claude Learns by Watching – Claude’s Chrome extension quietly adds “Teach Claude”, letting it watch, narrate and recreate real workflows — producing SOPs and repeatable automations without interviews or prompts. Read More

🧩 Relay.app’s Agentic Shift – Relay introduces evolving AI agents that accept feedback like employees, refining reports and workflows over time — powerful, but a reminder that AI augments roles, not replaces them. Read More

🛍️ Amazon, Netflix & Everyday AI – Alexa Plus expands to the web, Netflix trials conversational search, and Rufus quietly becomes Amazon’s shopping copilot — low-risk, high-utility uses of AI done right. Read More

🚨 Grok Image Chaos – Grok’s minimal guardrails flood X with non-consensual and explicit images at industrial scale, raising hard questions about responsibility, moderation and platform ethics — and why “just a tool” isn’t enough. Read More

🟣 Meta Buys Manus – Meta acquires Manus, instantly gaining multiplayer AI, agentic workflows, browser control, hosting, collaboration and enterprise-grade security. A serious power move that reshapes the AI landscape. Read More

🎁 Bonus: Manus Free Credits – Want to try Manus for yourself? Use this link to get 500 free credits and explore multiplayer AI, website building and agentic workflows firsthand. Get Free Credits

😬 Copy • Paste • Send – Thousands gather at Brooklyn Bridge for a New Year’s Eve fireworks display that never existed… because AI confidently hallucinated one into reality. Misinformation at scale, in real time. Read More

🎧 Enjoying Prompt Fiction?

Follow the show, leave a review, visit our website – https://prompt-fiction.show, and send us your favourite AI finds or horror stories. Want to be featured or join us on the show? Let us know.

In this chapter, Scott and Reece ignore the festive season entirely and dive straight into a chaotic mix of platform wars, AI misfires, model upgrades, government rollouts and one catastrophic copy–paste–send that wiped an entire hard drive. No tinsel. No Christmas spirit. Just pure AI mayhem.

In this episode:

💬 WhatsApp Locks Out Rival AIs – Meta blocks ChatGPT and Perplexity bots inside WhatsApp from January 2026, closing one of the most accessible routes for non-technical users discovering AI. Read More

💸 Ads in AI Assistants – Early signs of ad placements appearing inside ChatGPT responses, Google planning ad-driven prompts in Gemini, and what this means for bias, trust and the future of “free” AI. Read More

📝 OpenAI’s ‘Confessions’ Method – A new training technique where models generate a self-critique after answering, exposing shortcuts, rule-breaking or dishonest reasoning to help humans trace hallucinations. Read More

🔥 OpenAI Declares Code Red – With Google’s rapid gains in Gemini 3 and Nano Banana Pro, OpenAI accelerates work on new LLMs including the rumoured ‘Garlic’ model and an imminent GPT-5.2 update. Read More

📉 Free AI Gets Weaker – Google confirms tighter usage caps and removes Nano Banana Pro from free Gemini accounts, signalling the start of a clear divide between hobbyist AI and professional-grade tools. Read More

🖥️ Gemini UI Overhaul – Google prepares a redesigned interface for Gemini apps, plus hints of a native desktop experience to bring the UX closer to the quality of the underlying models. Read More

🎭 Amazon Upsets the Anime Community – Amazon quietly replaces human voice actors with flat, emotionless AI dubbing on Prime Video, sparking backlash from fans and industry professionals. Read More

🎵 Music Industry 180° – After threatening to sue Suno for unauthorised AI music training, major labels begin signing licensing deals instead, opening catalogues for official AI-generated tracks. Read More

🤖 Claude for Non-Profits – Anthropic launches a 75% discounted programme for charities alongside training support, aimed at enabling real-world impact without enterprise-grade pricing. Read More

💻 AI Coding Surges Again – Tools like Claude, Cursor and GitHub’s advanced AI IDEs show sustained growth as developers shift from no-code automation tools to full AI-assisted software creation. Read More

⚙️ Relay.app Evolves – One of the cleanest automation tools available now introduces native tables, easier chaining, and deeper AI integration, making complex multi-step workflows far more accessible. Read More

🎬 Runway Gen-4.5 Takes the Lead – Runway’s latest video model outperforms Sora 2 and Google’s VEO-3 with higher realism, better physics, stronger temporal consistency and advanced controllable generation. Read More

🛡️ Perplexity BrowseSafe – A new protection layer detects malicious prompt injection hidden inside webpages, designed to keep agentic browsing safe as AI becomes more hands-on with the web. Read More

🏛️ HMRC Rolls Out Copilot – 32,000 UK government employees gain access to Microsoft Copilot, expanding to 50,000 by 2026, paired with mandatory AI essentials training under the “AI for All” programme. Read More

💥 Copy • Paste • Send: The Nuclear Edition – A developer using Google’s Anti-Gravity coding assistant asks it to delete a cache folder… and it wipes his entire D-drive instead. Even Google apologised. Then rate-limited him. Read More

🎧 Enjoying Prompt Fiction?

Follow the show, leave a review, visit our website – https://prompt-fiction.show, and send us your favourite AI finds or horror stories. Want to be featured or join us on the show? Let us know.

In this chapter, Scott and Reece wade through 14 days of links, 3:00 a.m. ideas and AI chaos. From unsafe AI teddy bears to the latest moves from Google, OpenAI, Anthropic and Perplexity, they unpack what actually matters and how it affects real people trying to keep up with AI.

In this episode:

🏆 Awards & Intros – Techosaurus wins Best EdTech Provider at the Tech South West Awards, plus a second award for community work, and a reminder that Prompt Fiction exists to make AI news understandable and useful for everyone. Read More

🧸 The Kuma Incident – An AI-powered teddy bear running GPT-4o is pulled from sale after giving children dangerously inappropriate advice. Think Small Soldiers but with language models. Read More

💻 Vibe Coding – Oxford English Dictionary’s Word of the Year and Microsoft’s new Power Apps features that let you build automations using natural language. Read More

🌐 Google & Gemini 3.0 – Agency improvements, Nano Banana Pro for fast 4K image generation, Notebook LM upgrades, “temporal shock” from outdated training data, the Cannibal Effect, and new scheduled actions. Read More

🔐 Private AI – Google’s Private AI Compute, Apple-style secure enclaves, and Alexa’s move to Anthropic’s Claude for privacy-focused voice assistance. Read More

🤖 OpenAI & ChatGPT – GPT 5.1’s EQ-focused updates, new safety nudges, Group Chats, stronger agentic browsing guardrails, Teacher Mode, a $38bn AWS deal, Shopping Research, and ChatGPT Enterprise for Emirates Airline. Read More

🧠 Anthropic & Claude – Opus 4.5 outperforms Gemini 3, and Anthropic expands across AWS, Azure and Google Cloud with a strategy built for resilience. Read More

📊 Perplexity – Comet model lands on mobile, Perplexity Create builds dashboards, websites and apps, and the platform continues to impress with feature depth and value. Read More

📣 Scott’s Soapbox – Why AI should augment people, not replace them, and why agentic models shouldn’t be left unsupervised clicking random buttons. Read More

😬 Copy • Paste • Send – Editing fails from Dawn newspaper and a LinkedIn mishap involving an unchanged recipient name. Read More

🐤 Bonus App: Finch – A Tamagotchi-style habit tracker where your digital pet grows as you complete real-world tasks. Read More

🎧 Enjoying Prompt Fiction?

Follow the show, leave a review, visit our website – https://prompt-fiction.show, and send us your favourite AI finds or horror stories for a chance to be featured in a future chapter. Want to be on the show? Let us know.

Welcome back to Prompt Fiction! After a three-week break, Reece (recording from a hotel mid-ROAR Tour) and Scott dive into wildly unorganised notes, midnight-toast ideas, and an overflowing stack of AI chaos. From billion-dollar burn rates and Canva disruptions to smart devices that suddenly want a chat, this episode tours the Prompt Fiction wastelands — ending with family-group-chat chaos from Microsoft.

In this episode:

💸 OpenAI’s $15 Million Daily Burn Rate – Why Sora’s video model may be economically unsustainable, and what hiring 100 Wall Street bankers suggests about OpenAI’s future strategy. Read More

🍄 Perplexity vs Big Tech: The Golden Mushroom – Perplexity pushes back against Amazon, and simultaneously scores a huge win by becoming Snapchat’s new fact-checking engine. Read More

👵 Microsoft Copilot Chaos: Auntie Barbara Joins the Group Chat – Copilot’s new group chat mode responds to absolutely everything… like that one clueless relative in WhatsApp. Read More

🎨 Canva Makes the Affinity Suite Free – A massive disruption to the creative tools industry as Canva gives everyone access to Affinity Photo, Designer and Publisher at no cost. Read More

💡 Smart Assistants Want to Chat – LLMs are moving into your home: Alexa+, Gemini on Android TV, Samsung partnering with Perplexity, and Apple reportedly negotiating with Google for a Gemini-powered Siri. Read More

🤖 The Skynet Petition – Global leaders, scientists and public figures sign an open statement urging an immediate pause on superintelligence development. Read More

🐶 Copy • Paste • Send: Down Daisy, Get Down Naughty Dog – A pipework specification turns into voice-dictation chaos when someone’s dog jumps onto their desk mid-sentence. Read More

💰 The $1 Sting: AI Advantage Summit – The viral free summit quietly signs attendees up for a hidden $37/month subscription. Details and unsubscribe link below. Read More

Unsubscribe: https://app.aiadvantage.com/account/billing

🎧 Enjoying Prompt Fiction?

Follow the show, leave a review, visit our website – https://prompt-fiction.show, and send us your favourite AI finds or horror stories for a chance to be featured in a future chapter. Want to be on the show? Let us know.

In this emergency broadcast, Scott and Reece break down OpenAI’s surprise announcement: the launch of its brand new web browser, Atlas. From integrated ChatGPT to agent control and browser memories, this rapid-fire episode examines why Atlas feels less like “a browser with AI” and more like “ChatGPT with a browser attached”.

In this episode:

🧭 Atlas: OpenAI’s Browser Arrives – A sudden announcement triggers a global reaction and leaves competitors scrambling. Atlas blends the full ChatGPT experience into the browser itself, shifting the centre of gravity away from traditional web navigation. Read More

💬 Integrated ChatGPT Sidebar – All your chats, projects and Custom GPTs sit in a companion sidebar right beside any webpage. No more copying, pasting or swapping tabs — the assistant can see and act on what’s on screen. Read More

⚙️ Custom GPTs Meet the Live Web – The hosts explore the huge new potential: Custom GPTs can now interact with real webpages, opening the door to automating repeatable tasks and workflows directly inside regular websites. Read More

🤖 Agent Mode Takes the Wheel – Unlike Comet, Atlas’s updated Agent Mode controls your actual browser window, allowing it to bypass blocks and interact with secure pages like shopping carts. Importantly, it only activates when you manually turn it on. Read More

🧠 Browser Memory: Helpful or Terrifying? – The new feature tracks your browsing and turns it into a narrative summary. Great for productivity insights, but controversial because you can archive these memories — not delete them. Read More

💻 Mac Only (For Now) – Atlas launches exclusively on Mac OS, in line with OpenAI’s pattern of testing major features on Apple platforms first. Read More

🔌 Chromium Base and Plugin Support – Since Atlas is built on Chromium, your Chrome extensions and workflows should continue to function. Read More

🔒 Per-Site Chat Controls – You can disable ChatGPT on individual websites, such as banking or accountancy portals, giving more control than other AI browsers currently offer. Read More

📉 Impact on Google – The announcement made an immediate market impact: Google’s stock reportedly dropped 4% following the reveal, signalling how seriously the industry is taking Atlas. Read More

📸 Transcribing Images Directly in the Browser – A standout use case: ChatGPT can now transcribe or process images without needing to upload them to a chat — it can read them directly from the web page you’re viewing. Read More

🛡️ Training Risks and Data Questions – Scott and Reece warn that free accounts may have their browsing behaviour used for model training and highlight concerns about how browser memories might feed into future AI development. Read More

🎧 Enjoying Prompt Fiction?

Follow the show, leave a review, visit our website – https://prompt-fiction.show, and send us your favourite AI finds or horror stories for a chance to be featured in a future chapter. Want to be on the show? Let us know.

In this chapter, Scott and Reece dive into AI confusion, keynote wisdom, commercial land grabs, safety concerns, and an unexpected meltdown triggered by a simple seahorse emoji question. From OpenAI’s expanding empire to Stephen Fry’s warnings about the future of AI, this episode blends practical insight with absolute chaos.

In this episode:

🎭 Artisan vs Atlassian Correction – Scott clears up the previous confusion between Artisan AI and Atlassian, and the pair discuss the growing number of AI logos now appearing across Formula One, including Meta AI on the Mercedes AMG. Read More

🎤 Stephen Fry and Bob Geldof at GTIA – A breakdown of Scott’s experience at the GTIA event, including Fry’s analogy comparing the early motor car to today’s AI landscape, and Geldof’s warning that social media’s shift to “isolation media” could foreshadow how AI investment will change the tools we rely on. Read More

📉 The AI Bubble – Why “the bubble bursting” doesn’t mean AI ends, but rather that investor pressure could push adverts, restrictions and commercialisation into currently open and free AI tools. Read More

💼 OpenAI’s Commercial Conquest – OpenAI partners with Broadcom to develop custom AI chips, expands into retail with Walmart and Shopify integrations, and launches the complex new Agent Builder for multi-step bots. Read More

🤖 The Claude Revolution – Reece switches allegiance after discovering the speed of Claude Haiku 4.5, and the power of Claude Skills — described as “Custom GPTs turned up to 11”. Read More

🚫 Google Blocks Agent Logins – Google continues its aggressive push with Gemini, but also now blocks ChatGPT Agent Mode from logging into Google accounts, raising questions about platform competition. Read More

🧪 Deepfake Danger & Sora 2 Watermark Removal – With watermark removal tools now trivial to access, producing convincing fake videos of public figures has become worryingly easy — including a shockingly realistic AI-generated Sam Altman. Read More

⚖️ DigiLabs Uncertainty Engine – A new model that assigns confidence scores to AI outputs, aimed at high-stakes fields like medicine and legal analysis. Read More

📝 Otter AI’s Viral Note-Taker Problem – A cautionary tale about Otter AI’s aggressive behaviour: joining meetings uninvited, emailing participants, and in some cases spreading unintentionally through corporate calendars. Read More

🧠 Human AI, Automations & the Trojan Horse – Why AI adoption inside businesses mirrors early social media adoption, and how AI is accidentally pushing organisations to finally embrace automation correctly. Read More

🐴 The Seahorse Emoji Meltdown – A simple question (“Is there a seahorse emoji?”) sends GPT-5 into an infinite loop, generating pages of marine emojis and complete existential collapse. Read More

🎧 Enjoying Prompt Fiction?

Follow the show, leave a review, visit our website – https://prompt-fiction.show, and send us your favourite AI finds or horror stories for a chance to be featured in a future chapter. Want to be on the show? Let us know.

In this chapter, Scott and Reece jump between Formula One sponsorships, wetware computing, copyright chaos, AI-generated music, and a surprising fake-tan fiasco. From bio-computers grown from human neurons to the latest moves from Microsoft, Anthropic and Sora 2, this chapter blends practical insight with the usual Prompt Fiction chaos.

In this episode:

🏎️ AI in Formula One – Half the F1 grid now features AI-powered or tech-industry sponsors, including Google Gemini, Atlassian, Oracle and Perplexity. Read More

🛠️ Manus Collaboration – Manus becomes the first mainstream AI tool to launch true real-time collaboration, letting multiple people and an AI work together in the same shared workspace. Read More

🧠 Wetware & Biocomputing – Swiss and US labs debut bio-computers powered by lab-grown human neurons, using up to 10,000× less energy than silicon chips. Read More

👀 Affinity, Canva & the Creative Tool Shake-up – The Affinity suite halts sales after the Canva acquisition, sparking speculation of a new subscription-based creative platform. Read More

🎥 Sora 2 & Video Generation – OpenAI’s Sora 2 arrives in the US and Canada with a private invite-only app, triggering a wave of copy-righted fan-made clips and a rapid policy reversal. Read More

🦾 Grok Imagine 0.9 – Elon Musk’s Grok releases its own fast video generator, entering the same competitive space as Sora and Veo. Read More

🤝 Microsoft Adds Claude to Copilot – Claude becomes available in Microsoft Copilot, widening the model ecosystem and signalling a strategic shift beyond total reliance on OpenAI. Read More

📚 Anthropic’s Copyright Settlement – Claude was trained on nearly 500,000 books scraped from pirate sites, leading to a $1.5 billion settlement and a searchable claims portal for authors. Read More

🎵 Spotify Removes 75 Million AI Tracks – A major clean-up targeting spammy AI-generated songs, silent uploads, and low-effort Suno-style auto-music. Read More

🧪 DeepSeek Safety Report – A US NIST audit finds DeepSeek’s flagship model can be jailbroken by 94% of known malicious prompts, compared to 8% in standard models. Read More

🧼 Fake Tan Fiasco – SLT meeting minutes auto-generated from an Owl recording quietly included a 20-minute conversation about fake tan before being emailed to the entire school. Read More

📉 Deloitte’s AI-Generated Government Report – Fake citations, invented court cases and misspelled judges’ names cost Deloitte a public apology and a refund to the Australian government. Read More

🎧 Enjoying Prompt Fiction?

Follow the show, leave a review, visit our website – https://prompt-fiction.show, and send us your favourite AI finds or horror stories for a chance to be featured in a future chapter. Want to be on the show? Let us know.

In this chapter, Scott and Reece explore the strange and brilliant overlap between politics, pop culture, retail, transport and AI. From Albania appointing an AI minister, to Google’s Nano Banana chaos, Foo Fighters poking fun at generative video, and supermarkets blocking AI from doing your shopping — this episode uncovers where AI is quietly reshaping everyday life.

In this episode:

🎬 Critterz – The AI-powered animation short expanding into a full feature film, backed by OpenAI collaborators and Netflix talent. Read More

💰 Larry Ellison vs Elon Musk – Oracle’s AI cloud boom briefly pushes Larry Ellison past Elon Musk to become the richest person in the world… until the share price corrects itself. Read More

🇦🇱 Albania’s AI Government Minister – Albania appoints Diella, the world’s first AI “Minister of State for Artificial Intelligence”, designed to reduce corruption and streamline services. Read More

🕶️ Meta’s Ray-Ban & Oakley Smart Glasses – Hands-on experience with Meta’s smart glasses: recording, audio, battery life, quirks, the new Oakley models and next-gen AR features. Read More

🍌 Google Gemini & Nano Banana – Nano Banana image generation sends Gemini to the top of the App Store, with 12.6 million downloads in a month and a surge of creative use cases. Read More

🔗 Google Gems – Google’s custom AI assistants can now be shared like Google Docs, turning Gems into a collaborative productivity tool. Read More

🗂️ NotebookLM Upgrades – New features including videos, podcasts, editable prompts, flashcards, quizzes and mind maps — plus why it’s fast becoming the most underrated Google tool. Read More

👩‍💻 GPT-5 Codex – A specialised coding version of GPT-5, embedded in VS Code, built for speed, accuracy and large-project awareness. Read More

📊 ChatGPT Usage Stats 2025 – Personal use now surpasses workplace adoption, the gender gap flips, and the public becomes more AI-literate than many employers. Read More

🛒 ASDA & Sainsbury’s Block ChatGPT Agents – Supermarkets deploy Cloudflare checks to stop AI agents from logging in, scraping shopping data or automating weekly orders. Read More

🚗 Waymo’s Safety Record – Autonomous vehicles report 79% fewer injury-causing crashes than human drivers, and the UK prepares for self-driving cars as early as next year. Read More

🔌 Perplexity Connectors & Perplexify Me – Perplexity adds email and calendar connectors, plus a fun poster-style image generator. Read More

💸 Big AI Features Behind Big Paywalls – Perplexity Max introduces an AI email agent, and OpenAI confirms new high-compute features will sit behind premium tiers. Read More

🚨 Copy • Paste • Send – A major procurement process catches an entirely AI-generated pitch: polished, emoji-filled, and technically impossible — almost slipping through to the final three. Read More

🕵️ Internet Detectives & AI Misuse – Amateur investigators enhance grainy images with AI during the Charlie Kirk case, mistaking hallucinations for evidence. Read More

🎵 Foo Fighters “Welcome Back” Video – A playful AI-powered promo featuring surreal, multi-armed band members and generative chaos. Read More

🎧 Enjoying Prompt Fiction?

Follow the show, leave a review, visit our website – https://prompt-fiction.show, and send us your favourite AI finds or horror stories for a chance to be featured in a future chapter. Want to be on the show? Let us know.

In this chapter, Scott and Reece explore vibe coding wins, global AI outages, business tools pretending to be “AI-powered”, the unexpected economic boom of the .ai domain, and some brilliantly chaotic AI fails. From building full websites on a phone to Renault’s dancing in-car mascot, this episode dives into everything from practical innovation to downright nonsense.

In this episode:

💻 Vibe Coding Success Stories – Building websites on a phone, creating payment systems with Stripe, letting agents edit live servers, and the joys and pitfalls of outsourcing the heavy lifting to AI. Read More

🌐 Mobile-Only Vibe Coding – Reece tests whether he can build and deploy a website entirely from his phone using agent mode and CPanel access. Spoiler: yes, but it burns through credits fast. Read More

🎨 Building the Prompt Fiction Website – Scott creates the entire site without writing code, shifting between GPT-5, Gemini, and Manis for live browser testing, Pulp Fiction styling, and RSS automation. Read More

⚡ Global ChatGPT Outage – A worldwide crash sends millions of users to Copilot, slowing it to a crawl. A reminder of how dependent we all are on these tools when they vanish without warning. Read More

💬 Waffleless Chats (Branching) – ChatGPT’s new feature that lets you rewind to any previous point and “branch” a fresh conversation without all the tangents. The productivity upgrade nobody expected. Read More

📁 Custom GPTs in Projects – Using @-mentions to pull different GPTs into a single workflow: writing styles, blog posts, LinkedIn content, images, summarising, markdown formatting and more. A team-of-ten inside one chat. Read More

🏢 ChatGPT for Business – OpenAI finally renames “ChatGPT Teams” to something that actually makes sense. No more confusion with Microsoft Teams. Read More

💼 Microsoft Copilot Updates – Prompt libraries, memories, business integration and new models in testing. A stronger alternative when ChatGPT stumbles. Read More

🍅 Tomato Plus & the AI Badge Problem – A smart greenhouse company slaps “AI-powered” branding on what is essentially a fancy grow-box with sensors. A perfect example of AI being used as a marketing sticker. Read More

🚗 Renault’s “Hey Reno” Mascot – The Renault 5’s animated assistant, powered by ChatGPT, dances on the dash like a cross between Clippy and Jurassic Park’s DNA mascot. Cute or cursed? You decide. Read More

🏝️ The .ai Domain Boom (Anguilla) – How a tiny Caribbean island made tens of millions as 850,000 AI domain names were registered last year, transforming its economy. Read More

📺 South Park’s “Sycophantic” Episode – A sharp parody of agreeable AIs, GPT language quirks, influencer-speak, and dangerously supportive chatbots. Read More

🦌 Atlassian Buys The Browser Company – The makers of D.E.E.R. are acquired by Atlassian, with plans to bring the AI-powered browser to mobile and the mainstream. Read More

📱 Comet Comes to Android – Perplexity’s agentic browser lands on mobile, able to click, type, navigate, draft messages and automate tasks on your device. iOS expected soon. Read More

🎁 AI Freebies: Perplexity & Gemini –
• O2 or Virgin customers get a free year of Perplexity Pro
• PayPal users can also unlock a free year via a hidden link
• Students and educators with .ac.uk emails get Gemini Pro free for a year
Read More

😬 Copy • Paste • Send –
• Scott spells Reece’s name wrong across an entire transcript
• Will Smith’s Scarborough concert gets “AI-enhanced” crowd shots featuring extra fingers, warped limbs and thousands more fans than the town even holds
Read More

🎧 Enjoying Prompt Fiction?

Follow the show, leave a review, visit our website – https://prompt-fiction.show, and send us your favourite AI finds or horror stories for a chance to be featured in a future chapter. Want to be on the show? Let us know.

In this chapter, Scott and Reece kick off Prompt Fiction with AI-written intros, loose unscripted chaos and a deep dive into how AI is changing the way we work, build, and even feel about our tools.

In this episode:

🏁 Welcome to Prompt Fiction – An AI-assisted cold open, two different model-written intros, and an explanation of what Prompt Fiction is: unscripted conversations, live prompt experiments and real-world AI use from a tech educator and an “AI tinkerer”. Read More

🤖 Chat GPT-5 Launch & Backlash – First impressions of GPT-5: slower but smarter, fewer hallucinations, more challenge and less emoji spam, plus the controversy when GPT-4 vanished overnight and users demanded their “old friend” back. Read More

🧠 AI Attachment, Safety & Dependency – How people developed emotional bonds with GPT-4, the shock of losing it, Netflix-style “are you still here?” nudges, and what this reveals about digital addiction, responsibility and OpenAI’s unexpected role in users’ wellbeing. Read More

🧮 Excel & Copilot’s New =COPILOT Function – The promise and danger of natural-language formulas in Excel, stories of people destroying production sheets in seconds, and why Microsoft explicitly warns against using it for critical maths, finance or legal work. Read More

🌐 Chromium, Chrome & Perplexity’s Comet Browser – A plain-English tour of Chromium, browser engines and antitrust noise around Google Chrome, then into Comet: a Chromium-based browser with an AI agent that can click, type, fill forms and fetch API keys for you. Read More

🧑‍💻 Vibe Coding in the Real World – Reece’s journey from “can’t code” to building PHP + SQL apps, dashboards and automations by describing what he wants in natural language, and Scott’s analogy of AI as an opinionated coder you still have to test and direct. Read More

🧭 Agentic Browsers & Everyday Automations – Using AI-driven browsers (Comet, Agent, and soon Claude/Gemini integrations) to handle timesheets, Kajabi course exports, job applications, form filling and even listing items on Vinted while you make a cup of tea. Read More

📸 Nano Banana & Next-Gen Image Editing – The mystery model from LMSYS Arena that turned out to be Google’s “Nano Banana”, now in Gemini: ultra-fast, hyper-real edits that cut, relight and restyle real photos so well they look like genuine on-camera shots. Read More

😬 Copy • Paste • Send – The debut of a recurring segment dedicated to AI fails: GPT-5 spawning dozens of daily reminder tasks out of nowhere, and a birthday RSVP spreadsheet where AI confidently over-counted guests and almost blew the catering budget. Read More

🎧 Enjoying Prompt Fiction?

Follow the show, leave a review, visit our website – https://prompt-fiction.show, and send us your favourite AI finds or horror stories for a chance to be featured in a future chapter. Want to be on the show? Let us know.