LinkedIn CringeBot 3000
• LinkedIn CringeBot 3000 is an AI tool that transforms topics into typical LinkedIn-style posts.Read more | Comments | @hackernws
• Users input a topic and select from six cringe styles like 'The Hot Take' and 'The Vulnerability Post'.
• The tool generates authentic thought leadership™ content mimicking current LinkedIn trends.
• Styles include empowerment, shameless promotion, quitting 9-5, and speaking uncomfortable truths.
• Website launched on August 11, 2026; offers sponsorship via email cringebot3000@proton.me.#LinkedInCringe #AIWriting #SocialMediaSatire
Cringebot3000
LinkedIn CringeBot 3000
Transform any topic into peak LinkedIn thought leadership guaranteed to make your followers shudder.
Facebook is paying controversial creators to produce rage-bait content
• Far-right agitator Hugo Lennon monetized on Facebook since Sept 2025, despite racist content.Read more | Comments | @hackernws
• The Noticer, promoting neo-Nazi views, earned from Facebook's program since Nov 2025.
• Anti-immigration and anti-vaccine pages also paid by Facebook's content monetization program.
• Facebook claims monetization policies are enforced, but controversial creators remain funded.
• Meta paid nearly $4.27 billion to 16.2 million accounts in 2025, with some policy violations noted.#FacebookMonetization #FarRightContent #MetaTransparency
www.abc.net.au
Facebook pays controversial creators to produce rage-bait content
An ABC NEWS Verify investigation reveals Facebook is directly funding some controversial Australian content creators including a white nationalist and an anti-vaxxer influencer.
Tech leaders say AI means less work – staff say they work up to 90 hours a week
• Tech executives predict AI will reduce work hours, aiming for four-day work weeks by 2025.Read more | Comments | @hackernws
• Former OpenAI employee reports 70-90 hour work weeks during sprints, with weekend work common.
• Meta employees report being involuntarily reassigned to AI projects with long, on-call hours.
• UC Berkeley study finds AI increases workload pace, task breadth, and work hours for tech workers.
• AI time savings often absorbed by additional tasks and constant output checking, negating workload reduction.#AIWorkload #TechIndustry #WorkHours
Bbc
Tech leaders say AI means less work - their staff say they work up to 90 hours a week
Tech companies are not modelling their own claims of the technology giving people more free time.
US hires over 2k video gamers as air traffic controllers
• FAA recruited over 2,000 video gamers to train as air traffic controllers.Read more | Comments | @hackernws
• Gamers possess skills like quick thinking and multitasking relevant to air traffic control.
• FAA meets 94% of its hiring goal with more than 2,000 candidates in pipeline.
• Median air traffic controller salary was $144,580 in 2024.
• Candidates must pass FAA tests, be US citizens, under 31, and pass an aptitude exam.#AirTrafficControl #GamingSkills #FAARecruitment
CBS NEWS
Play video games? You, too, could become an air traffic controller.
Transportation Secretary Sean Duffy is touting the success of a campaign targeting video gamers to train as air traffic controllers.
The Human Is the Loop
• Author took a multi-week break from AI usage during summer 2026.Read more | Comments | @hackernws
• Returned to find many unfinished AI tasks causing guilt and stress.
• Recognized AI use often acted as a crutch, reducing curiosity and confidence.
• Noted AI fosters dependency and complex productivity loops, not always beneficial.
• Plans intentional AI use to enhance work without losing human insight and control.#AIUsage #HumanInTheLoop #TechHabits
Brent Fitzgerald
The human is the loop
After a few weeks away from AI, I see my usage wasn't always productive or healthy.
2026 Eclipse Webcams
• The next total solar eclipse will occur in 2026.Read more | Comments | @hackernws
• A website has interactive webcams tracking the eclipse path.
• Totality begins in approximately 3 hours 34 minutes from the snapshot time.
• The eclipse will reach the first webcam in about 4 hours 14 minutes.
• The site uses Leaflet and OpenStreetMap for map visualization.#SolarEclipse2026 #TotalEclipse #LiveWebcams
What sort of maths are LLMs good at?
• In August 2026, OpenAI's LLMs solved 10 major math problems, including a non-sofic group and a superexponential Ramsey number lower bound.Read more | Comments | @hackernws
• LLMs excel at finding counterexamples and examples, often via brute-force trial and broad mathematical knowledge.
• Finding a counterexample depends on context; existential statements can be examples or counterexamples.
• LLMs likely rely on exploring vast search trees, lacking humans' ability to prune unfruitful paths efficiently.
• Future breakthroughs depend on LLMs developing 'mathematical nose' for novel, deep insights beyond brute force.#AIinMath #MathBreakthrough #LLM
Gowers's Weblog
What sort of maths are LLMs good at?
For the sake of anyone who might read this blog post in the distant future (a month from now, say), let me mention that I am writing it a few days after OpenAI announced that it had solved ten majo…
Show HN: Woxi - Open-source Mathematica / Wolfram Language reimplementation
• Woxi is a Wolfram Language interpreter implemented in Rust.Read more | Comments | @hackernws
• Runs entirely in the browser locally without sending data externally.
• Available as playground, command-line tool, Jupyter kernel, and native notebook editor.
• Supports graphical outputs like plots, charts, graphs, and tree forms.
• Woxi Studio edits .nb files and exports to formats like .ipynb, Markdown, and PDF.#WolframLanguage #RustLang #WebAssembly
Woxi
Woxi — Wolfram Language interpreter in Rust
Run Wolfram Language code in the browser playground,
on the CLI, in Jupyter, or in the Woxi Studio notebook editor.
on the CLI, in Jupyter, or in the Woxi Studio notebook editor.
AI is removing the middle class of software engineering
• AI enables rapid, large-scale code changes, increasing complexity and technical debt.Read more | Comments | @hackernws
• Teams struggle to understand and maintain AI-generated code, leading to convoluted systems.
• Bad engineering decisions compound faster due to AI's speed, hard to reverse.
• High-value engineers are those who can effectively judge and manage AI contributions.
• AI polarizes engineering roles, making mediocre skills less employable and top skills more valuable.#AIinEngineering #TechIndustry #SoftwareDevelopment
Blog - Florian Herrengt
AI is removing the middle class of software engineering
AI makes projects with weak engineering culture fail much faster.
License Plate Reader Searches Should Require a Warrant
• Norfolk court ruled ALPR data collection not an illegal search, but future warrant likely needed.Read more | Comments | @hackernws
• ALPR cameras cost under $3,000 each, effective for crime flagging and active alerts.
• Historical ALPR searches (past 3 minutes) should require police warrant to prevent abuse.
• Data deletion policies do not prevent misuse, limit law enforcement and defense utility.
• States should mandate warrants and independent audits for historical ALPR data searches.#ALPR #PrivacyLaw #PoliceSurveillance
Andrew Wheeler
License Plate Reader Searches Should Require a Warrant
So while I work with police departments regularly, I think it is critically important that technology be used reasonably. While this may be off-putting to some of my clients, I worked with the Inst…
Tailscale Traces Database Corruption to 16y/o SQLite WAL-Reset Bug
• Tailscale experienced 19 SQLite database corruption incidents over six months starting August 2025.Read more | Comments | @hackernws
• The corruption was caused by a 16-year-old SQLite bug called the WAL-Reset bug involving a checkpoint-write race.
• The bug was fixed in SQLite 3.52.0, though an associated false alarm issue led to a revised 3.51.3 release.
• Tailscale uses manual checkpointing of SQLite WAL files and aggressive backups, triggering the rare bug.
• After deploying the fix, Tailscale confirmed the bug's activity via warnings and stabilized their database reliability.#SQLiteBug #DatabaseCorruption #TailscaleFix
Tailscale
How Tailscale helped find the SQLite WAL-Reset bug
Tailscale and SQLite developers traced maddening corruption incidents to find the WAL-Reset data race, then uncovered a second stale expression index bug.
Facebook ads are so hard to block that uBlock Origin stopped filtering them
• uBlock Origin developers announced they will stop filtering Facebook ads due to complexity.Read more | Comments | @hackernws
• Facebook uses tactics like splitting and randomizing 'Sponsored' tag letters to evade filters.
• Facebook's ad blocking resistance has lasted about five years with evolving anti-blocking methods.
• Google Chrome's switch to Manifest V3 has limited uBlock Origin's effectiveness.
• Firefox and Brave browsers still support uBlock Origin, but Chrome and Edge phase it out.#FacebookAds #uBlockOrigin #AdBlocking
Neowin
Facebook ads are so hard to block that uBlock Origin stopped filtering them
Facebook ads and ad-blocking tools have been in a constant clash for years. But it looks like uBlock Origin is stepping away from that fight.
Worms: The Future of Yesterday's Worms Today
• Worms: The Director's Cut - 30th Anniversary Expansion includes 40+ new levels and 50+ sample packs.Read more | Comments | @hackernws
• The expansion adds 10 DIY terrains and many new teams to the game.
• A remix album with nine new tracks inspired by the original Worms theme is available for download.
• An online tool allows users to create dual-layer maps, mountain sets, terrains, and teams via browser.
• The project was curated and produced by Andy Davidson and Ben 'Squirminator2k' Paddon, under Team17 Digital Ltd.#Worms30th #GamingExpansion #Team17
Qwen/Qwen3.8-2.4T-A95B
• Qwen3.8-2.4T-A95B is a 2.4 trillion parameter causal language model with 95B activated parameters.Read more | Comments | @hackernws
• It supports a native context length of 262,144 tokens, extendable to over 1 million tokens.
• Qwen3.8 enhances coding, professional work, research, and long-horizon task capabilities over previous Qwen versions.
• It integrates with SGLang, vLLM, TokenSpeed and offers APIs via Qwen Cloud for scalable inference.
• Benchmark tests show Qwen3.8-Max generally outperforms Qwen3.7-Max and approaches or exceeds scores of GPT-5.6 Sol on many coding and general benchmarks.#Qwen38 #AIModel #LanguageModel
huggingface.co
Qwen/Qwen3.8-2.4T-A95B · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
DeepSeek V4 Pro 0813
• DeepSeek V4 Pro 0813 is a large-scale mixture-of-experts AI model released on Aug 12, 2026.Read more | Comments | @hackernws
• Pricing is $0.435 per million input tokens and $0.87 per million output tokens with 81.25% cache hit rate.
• The model supports a 1 million token context window and reasoning-enabled features.
• Performance: 51 tokens/s throughput with 1.26s median latency and 100% uptime over 3 days.
• Hosted solely by DeepSeek provider, used in apps like Velocity CLI and Hermes Agent handling tens of millions of tokens.#DeepSeek #AIModel #OpenRouter
openrouter.ai
DeepSeek V4 Pro 0813 - API Pricing & Benchmarks
DeepSeek V4 Pro 0813 is a large-scale mixture-of-experts model from DeepSeek. $0.435 per million input tokens, $0.87 per million output tokens. 1,048,576 token context window, maximum output of 384,000 tokens. Higher uptime with 5 providers. Includes independent…
Why Tiny JPEGs Look Different in Chrome
• Chrome uses partial IDCT scaling to efficiently decode JPEGs at small sizes by skipping high-frequency data.Read more | Comments | @hackernws
• JPEG images are compressed in 8x8 blocks via DCT, storing frequency coefficients from low to high.
• Rendering a heavily downscaled JPEG keeps mainly low-frequency information, losing fine details.
• Chrome's Skia and libjpeg-turbo decode at closest 1/8 fraction scale, then downsample to size.
• JPEG is suboptimal for icons due to its photo-focused compression and decoding method.#JPEGDecoding #ChromeOptimization #ImageScaling
guillaumetech.github.io
Guillaume Técher
Guillaume Técher — Personal website
Grok 4.6
• Grok 4.6 matches GPT-5.6 Sol on the Artificial Analysis Intelligence Index.Read more | Comments | @hackernws
• It features improved training with more data, better optimizer, and longer runs than Grok 4.5.
• Grok 4.6 excels at multi-step projects, converting ideas into functional apps with iterative feedback.
• Available now in Cursor and Grok Build platforms with 2x free usage in the first week.
• Pricing starts at $2/million input tokens and $6/million output tokens; fast variant costs twice as much.#AI #Grok46 #MachineLearning
x.ai
Introducing Grok 4.6
Grok 4.6 builds on Grok 4.5 with a particular focus on long-running agents and more ambitious interactive and visual work.
uBlock Origin Is Giving Up the Fight to Keep Ads Off Facebook
• The article from digitalescapetools.com is inaccessible due to a 403 Forbidden error.Read more | Comments | @hackernws
• It likely discusses uBlock Origin's change in handling Facebook ads.
• JavaScript and cookies need enabling to view content on the site.#uBlockOrigin #FacebookAds #AdBlocking
Digitalescapetools
uBlock Origin Is Giving Up the Fight to Keep Ads Off Facebook
uBlock Origin's volunteer team will no longer chase Facebook's shifting ad-detection tricks, calling it a 'disgusting anti-user site.' Here's what that means for your feed.
A shell exclamation mark is not for yelling. Be lazy
• Shell event designators using '!' allow referencing and modifying previous commands and arguments.Read more | Comments | @hackernws
• Key designators include !! (previous command), !string (last command starting with string), !?$ (last argument).
• Modifiers like :h, :t, :r, :e extract parts of arguments (dirname, basename, extension).
• fc command edits previous commands in your editor and can re-run or substitute text in them.
• bash and zsh can verify expansions with histverify/HIST_VERIFY to prevent unwanted execution of expansions.#ShellTips #CommandLineMagic #Bash
refp.se
A shell exclamation mark is not for yelling. Be lazy. | Filip Roséen - refp.se
Event designators have been hiding in not-so-plain-sight since the late 1970s — so powerful that forgetting might be part of a massive conspiracy to wear out developer keyboards faster than otherwise.
SpaceXAI's Grok 4.6 Scores 61 on the Artificial Analysis Intelligence Index
• SpaceXAI's Grok 4.6 scores 61 on Artificial Analysis Intelligence Index, matching GPT-5.6 Sol.Read more | Comments | @hackernws
• Grok 4.6 exceeds Grok 4.5 by 5 points and Grok 4.3 by 23 points on Intelligence Index.
• Pricing remains $2/$6 per 1M tokens, 60%+ cheaper than Claude Opus 5 and GPT-5.6 Sol.
• Grok 4.6 shows strong agentic performance with GDPval-AA v2 Elo of 1753, near top models.
• It completes long-horizon tasks efficiently in ~53 turns and 0.5B tokens, half of the top competitor's usage.#AIFrontier #Grok46 #TechInnovation
artificialanalysis.ai
Grok 4.6 returns SpaceXAI to the intelligence frontier and leads on cost efficiency
SpaceXAI returns to the intelligence frontier with strengths in agentic performance and cost efficiency
Someone is running mass vulnerability scans, spoofing AI bots like ClaudeBot
• 35% of website visits come from bots, down 2% over last 90 days.Read more | Comments | @hackernws
• 28% of bot traffic is AI-related, up 11% in the last 90 days.
• 98.5% compliance rate of bots following robots.txt rules.
• Top AI scrapers: ClaudeBot (27%), meta-externalagent (19.5%), Amazonbot (19.2%).
• ChatGPT-User accounts for 86.4% of AI fetching traffic for real-time content.#AIBotTraffic #WebAnalytics #RobotsTxt
Known Agents
The Agentic Web Index: AI Bot Traffic Statistics | Known Agents
A continuously updating analysis of bot vs. human traffic, AI scraping, fetching, search indexing, browsing, robots.txt compliance, and AI chat referrals across 5,000+ websites.