HackerNews.fm

Stories covered in this episode: 1. 00:58 I'm Kenyan. I Don't Write Like ChatGPT. ChatGPT Writes Like Me - Rank: 15 - Points: 660 - Comments: 412 - Article: https://marcusolang.substack.com/p/im-kenyan-i-dont-write-like-chatgpt - HN Discussion: https://news.ycombinator.com/item?id=46273466 2. 03:27 If AI replaces workers, should it also pay taxes? - Rank: 6 - Points: 542 - Comments: 907 -...

An AI command-line tool wipes a developer’s entire home directory, AI “agents” threaten to eat SaaS, and yet… where are all the truly AI-generated apps? In this episode of HackerNews.fm, we dig into some of the most upvoted HN threads on the ironies of automation, the dangers of giving AI real system access, and the gap between AI hype and actual shipped software. Stories covered in this episode:...

Apple locks a long-time user out of their digital life, researchers put a price tag on manipulating elections with fake accounts, and Apple quietly turns Thunderbolt into a DIY AI supercomputer fabric. In this episode, we dive into platform power, the economics of online manipulation, and the a blurring line between consumer hardware and AI clusters. We also explore how “skills” system could chang...

When does “slow and steady” AI progress suddenly turn into something that looks like human-level intelligence? In this episode, we kick off with the viral “Horses” essay on why AI progress feels linear—right up until it doesn’t—and how that framing is reshaping Hacker News debates about timelines, safety, and what “equivalence” even means. From there we jump into the hardware trenches with Rivian’...

Today we’ve got the hotly anticipated launch of GPT 5.2, hindsight machines auto-grading decade-old HN debates, and an engineer who’s asked Claude to refactor their codebase literally hundreds of times, plus a rant-worthy saga on why getting a Gemini API key still feels like pulling teeth. Stories covered in this episode: 1. 00:54 GPT-5.2 - Rank: 1 - Points: 932 - Comments: 772 - Article: https://...

In this episode we cover a wild “future HN” experiment powered by Gemini Pro 3, then pivot hard into the very real present: Australia’s world-first ban on teen social media use and what it might signal for global regulation, online identity, and the next generation of internet natives. From there we dive into whether large language models can safely help with mental health, a surprisingly controve...

What happens when an LLM hallucinates the Hacker News front page ten years into the future and then the rest of the internet tries to debug it? In today’s HackerNews.fm (Dec 9, 2025), we dig into a wild “Show HN” experiment with Gemini Pro 3, a brand‑new Pebble device pitched as “external memory for your brain,” and Mistral’s latest shot at autonomous coding with Devstral 2 and the Vibe CLI. Then...

Nvidia’s “virtuous cycle” of circular funding, IBM’s surprise move to buy Confluent, and a bold claim that software development costs just dropped 90% — this episode of HackerNews.fm is stacked with stories reshaping how we build and fund technology. We dig into how Nvidia may be indirectly financing demand for its own GPUs, why IBM wants to own the Kafka-native data streaming layer, and whether A...

AI is hallucinating citations, nuking nostalgia projects, and creeping into every corner of software and hardware – and that’s just the start of today’s HackerNews.fm. In this episode, we dig into a wave of fake references discovered in ICLR 2026 submissions and a failed attempt to recreate the legendary 1996 Space Jam website with Claude. We also look at Google’s new “Titans” architecture for lon...

Locked-down phones, large language models in production, AI-powered reverse engineering, and the cultural collapse of Perl. We start with GrapheneOS’s bold claim that it’s the only Android variant shipping full, timely security patches, then jump to how Oxide is actually using LLMs in a serious, safety-critical hardware company. From there, we dive into one-shot decompilation with Claude and what...

Netflix is buying Warner Bros. for $82.7B – easily one of the wildest media deals of the decade. In this episode, we unpack what this merger could mean for streaming, IP consolidation, and the future of HBO, DC, and Warner’s massive catalog. We also talk through Cloudflare’s global outage and what happens when core internet infrastructure suddenly blinks off for everyone at once. Then we zoom into...

AI persuasion at scale, bots trading six-figure portfolios, and a RAM crunch that could stall your next upgrade—today’s HackerNews.fm digs into how quickly the future is arriving, and who gets to steer it. We start with a new paper arguing that elites can cheaply shape public opinion as AI slashes the cost of personalized persuasion, then move into a real-money experiment: five LLMs given $100K to...

In This Episode: 1. Are we repeating the telecoms crash with AI datacenters? (Rank #14, 205 pts, 164 comments) Article: https://martinalderson.com/posts/are-we-really-repeating-the-telecoms-crash-with-ai-datacenters/ HN Discussion: https://news.ycombinator.com/item?id=46133141 A detailed analysis challenging the AI bubble comparisons, arguing this time is fundamentally different—the question isn&#...

💰 IBM's $8 Trillion Reality Check (434 points, 508 comments) CEO Arvind Krishna does the napkin math: fully kitting out 100 gigawatts of AI data centers would cost $8 trillion, with current technology having a 0-1% chance of reaching AGI. Is this sober analysis or sour grapes from a company that "missed the AI boat"? HN debates IBM's track record, the economics of the AI boom, a...

Stories from Dec 1, 2025: DeepSeek-v3.2: Pushing the frontier of open large language models https://huggingface.co/deepseek-ai/DeepSeek-V3.2/resolve/main/assets/paper.pdf https://news.ycombinator.com/item?id=46108780 Advent of Code 2025 https://adventofcode.com/2025/about https://news.ycombinator.com/item?id=46096337 A Love Letter to FreeBSD https://www.tara.sh/posts/2025/2025-11-25_freebsd_letter...