HN Flash

Hacker News. Daily summary. Top 20 stories.

2026-10-28

Ghostty 1.4 (Ghostty)

Ghostty 1.4 adds full Kitty graphics support, major performance and memory improvements.

Mixed sentiment: some question the need for new terminals, others value cross-platform consistency and speed.

1156 pts · 366 comments

US Government is hiding a secret court ruling about how it spies on Americans (WIRED)

The Trump administration is accused of illegally withholding a FISA court ruling on surveillance.

Secret courts are widely condemned as undemocratic, with some arguing both parties support surveillance.

1017 pts · 388 comments

Genie 4 (deepmind.google)

DeepMind's Genie 4 generates real-time interactive 3D worlds with audio and speech.

Skepticism dominates: commenters doubt the demo's interactivity and warn of a Stadia-like DRM future.

957 pts · 548 comments

AI promised less work. I've never worked more (oioannou.com)

An engineer finds AI agents let him do more but leave him exhausted from constant context-switching.

Shared burnout from AI's treadmill: speeding up the machine but trapping humans in constant context-switching.

651 pts · 646 comments

Python's experimental JIT to be removed (LWN.net)

Python's experimental JIT compiler is being removed from CPython after PEP 836 was withdrawn.

Two camps: those frustrated by bureaucratic barriers to performance gains, and those defending the council's caution on a major feature.

545 pts · 298 comments

Anthropic trained Claude on illegally obtained Stanford professors' books (The Stanford Daily)

Judge ruled Anthropic's AI training on legally acquired books is fair use, but pirating them is illegal.

Thread splits: training is fair use, but pirating the source is the crime.

541 pts · 280 comments

Wikipedia's Cleanup of AI-Generated Content (IEEE Spectrum)

Wikipedia editors have removed an estimated 15-20% of new articles as AI-generated as of early 2026.

Split between those who see LLMs as inherently unreliable for encyclopedic content and those who think they could work with proper verification tools.

484 pts · 222 comments

Roblox downloaded 145GB models from us (including adult content) without asking (Open3DLab)

Roblox's AI crawler downloaded 145 GB of 3D models from Open3DLab, including adult content, without permission.

Technical advice on blocking datacenter IPs dominated, alongside skepticism about the adult-content angle.

477 pts · 270 comments

The Mythical Man-Month in the Age of AI (Jan van den Berg)

Brooks's Law still applies to AI agents, only the bottlenecks have changed.

Agents produce wrong answers faster, which some see as a feature for requirements discovery.

462 pts · 335 comments

Survival LLM: AI assistant and emergency library that runs on your own computer (brangus.io)

Survival LLM is a free offline AI assistant with an emergency library for your computer.

Debate over whether an LLM is appropriate for survival tools versus a simpler searchable archive.

422 pts · 159 comments

New 14‑inch MacBook Pro with M6 (Apple Newsroom (India))

Apple announces the 14-inch MacBook Pro with M6, featuring up to 30% faster graphics than M5.

Many M1 users see no reason to upgrade; debate over battery life claims versus Intel competition.

389 pts · 535 comments

Police are using AI to pull over drivers for 'suspicious' behavior (gizmodo.com)

ACLU warns AI-powered license plate readers enable police to flag drivers for 'suspicious' travel patterns.

Commenters overwhelmingly condemn the surveillance as unconstitutional and authoritarian, with some calling for federal bans.

379 pts · 299 comments

Ten days later, Git on macOS still trusts the compromised eMudhra certificates (we-have-a-problem.com)

Apple's git on macOS still trusts compromised eMudhra certificates ten days after notification.

Commenters fault Apple for not patching macOS CLI tools after 11 days.

360 pts · 146 comments

New iMac with M6 (Apple)

Apple released an iMac with an M6 chip, 24-inch display, and 60Hz refresh rate.

Two camps: those defending 60Hz as fine for the target audience, and those arguing for higher refresh rates to reduce eye strain.

353 pts · 447 comments

Git for the US Code (Lexifina)

A project applies Git-like version control to the U.S. Code, tracking legal text changes.

Prior work by the Library of Congress and commercial legal databases already tackles this.

343 pts · 178 comments

Amazon Is Gutting Its AI Division After Sustained Failure (Futurism)

Amazon is gutting its AI division after its Nova models failed.

Amazon's risk-averse culture and data restrictions are seen as barriers to AI innovation.

333 pts · 262 comments

Software may be getting worse, but it used to be horrible (www.bitlog.com)

The author argues software quality has improved dramatically since the 1990s, despite current complaints.

Two camps: reliability is vastly improved, but UX and user control have declined.

331 pts · 337 comments

Tailscale SSH: Investigation and remediation update (tailscale.com)

Tailscale discloses privilege escalation vulnerabilities in its SSH feature and plans a patch by October 27.

Commenters split between those questioning the severity and those advocating removal of Tailscale due to growing attack surface.

323 pts · 170 comments

Discord will soon let anyone find public server messages through Google (Discord)

Discord will let public server messages appear in Google searches starting mid-December 2026.

Mixed reactions: some welcome searchable Discord content, others push open alternatives like Fluxer.

312 pts · 188 comments

ICE Is Running One of the Largest Anti-Left Surveillance Ops in US History (WIRED)

ICE is running one of the largest anti-left surveillance operations in US history, according to court filings.

Alarm over surveillance and police state claims; split on Second Amendment relevance and severity of charges.

304 pts · 115 comments