惯性聚合 高效追踪和阅读你感兴趣的博客、新闻、科技资讯
阅读原文 在惯性聚合中打开

推荐订阅源

H
Hackread – Cybersecurity News, Data Breaches, AI and More
让小产品的独立变现更简单 - ezindie.com
让小产品的独立变现更简单 - ezindie.com
V
V2EX
T
The Blog of Author Tim Ferriss
腾讯CDC
Hugging Face - Blog
Hugging Face - Blog
雷峰网
雷峰网
爱范儿
爱范儿
GbyAI
GbyAI
H
Help Net Security
I
InfoQ
罗磊的独立博客
酷 壳 – CoolShell
酷 壳 – CoolShell
钛媒体:引领未来商业与生活新知
钛媒体:引领未来商业与生活新知
cs.AI updates on arXiv.org
cs.AI updates on arXiv.org
人人都是产品经理
人人都是产品经理
J
Java Code Geeks
Microsoft Security Blog
Microsoft Security Blog
Threat Intelligence Blog | Flashpoint
Threat Intelligence Blog | Flashpoint
N
Netflix TechBlog - Medium
Last Week in AI
Last Week in AI
宝玉的分享
宝玉的分享
云风的 BLOG
云风的 BLOG
Project Zero
Project Zero
P
Privacy & Cybersecurity Law Blog
A
Arctic Wolf
Know Your Adversary
Know Your Adversary
G
Google Developers Blog
freeCodeCamp Programming Tutorials: Python, JavaScript, Git & More
T
Tor Project blog
V
Vulnerabilities – Threatpost
Y
Y Combinator Blog
WordPress大学
WordPress大学
V
Visual Studio Blog
博客园_首页
G
GRAHAM CLULEY
K
Kaspersky official blog
T
Tailwind CSS Blog
T
Threat Research - Cisco Blogs
博客园 - Franky
D
Docker
Security Latest
Security Latest
I
Intezer
有赞技术团队
有赞技术团队
Application and Cybersecurity Blog
Application and Cybersecurity Blog
博客园 - 【当耐特】
B
Blog RSS Feed
T
The Exploit Database - CXSecurity.com
cs.CV updates on arXiv.org
cs.CV updates on arXiv.org
OSCHINA 社区最新新闻
OSCHINA 社区最新新闻

404 Media

We Are Living in a ‘ChatGPT Flyer Pandemic’ 'Knockoff' Browser Extension Hides Sketchy Brands on Amazon LARPING: How Influencers Fake Being Rich Cops Say Waymo Snitched on Teens for Allegedly Drinking and Shooting a Toy Gun Scientists Gave Mice Cocaine. This Is What It Did to Their Brains Footage Shows Cop Stalking Woman He Met on a TV Set After Surveilling Her With a License Plate Reader SOLVED: The Case of the Missing Megalodon Behind the Blog: With Blogs Like These, Who Needs a Private Jet Companies Are Throttling Employees’ AI Use Because It’s Too Expensive Podcast: The AI Tokenpocalypse Is Here Scientists Asked AI to Impersonate 112 Public Figures. What Happened Next Is a ‘Dire’ Warning Apple ‘Hide My Email’ Vulnerability Reveals Peoples’ Real Email Addresses I Have Thoughts About That Kylie Jenner Meta Glasses Ad County With 37 Data Centers Asks Schools to ‘Conserve Electricity’ Scammers Sell Seeds for Exotic AI-Generated Flowers That Don’t Exist How I Bought a Private Jet By Selling $10 Subscriptions to 404 Media Tidal Says It Won’t Pay Royalties for AI-Generated Music Inside Cannes, the Advertising Industry’s Biggest Party Scientists Think They’ve Uncovered the 15-Million-Year-Old Origin of Laughter Behind the Blog: Salesforce Beach Bodycam Shows Moment Cops Arrested a Man for Speaking Too Long at Data Center Meeting Vast ‘Structures’ In Space Reveal the Universe Isn't What We Thought The Trump Administration’s New Census Data Rules Are a Policy Disaster Snap's AI Specs: LOL The Tokenpocalypse Is Here: Companies Are Scrambling To Stop Spending So Much on AI Podcast: If AI Is Sentient Then So Is ‘Age of Empires II How Hackers Broke into Madison Square Garden Wikipedia Cofounder Larry Sanger Banned From Site for ‘Canvassing’ Libraries Not Doing Pride Displays Say They ‘Shouldn’t Be Judged’ Madison Square Garden Made Dossier on Activists Who Opposed Facial Recognition 'We Will Fight to Our Very Last Breath:' Township Leaders Vow to Fight Nuclear AI Data Center Stopping Tech Company Censorship (with Jake Hanrahan) Are Public Libraries Becoming Children’s Libraries? Scientists Propose Black Holes Don’t Exist, Are Something Much Stranger Behind the Blog: Landfillcore and Go Knicks A New Fossil Discovery Just Rewrote 150 Years of Evolutionary Theory If AI Is Sentient Then So Is ‘Age of Empires II’ Salesforce’s Internal AI Leaderboard Has Teams Competing for Little Trophies ICE Appears to Be Buying Immigrants’ Tax Identifiers from a Data Broker Podcast: The Government Wants to End Anonymity on Phones Hackers Publish Knicks and Madison Square Garden Data Online Hackers Are Hijacking Entire Roblox Games Now Disclosure Day's Delusion Is That People Would Think Alien Videos Are Not AI It Is Trivially Easy to Use Reddit to Manipulate AI Search, Research Suggests Judge Rules Blacked.com Can Sue Meta for Scraping Its Porn The OPSEC Rave Wave (with Imani Thompson) Scientists Discover Vast Ancient ‘Necropolis’ Teeming With Strange New Creatures ‘You Will Not Speak on Flock Tonight’: County Commissioner Refuses to Let Residents Opposing Flock Speak at Meeting Behind the Blog: World Cup Madness and Film Reviews Software Update Automatically Turns off Amazon Delivery Drivers’ AC During Dangerous Summer Heat Amazon Data Centers In Mississippi Have Already Raised Electricity Rates for Local Customers, Report Suggests Flock Leaked Cops’ License Plate Searches via DuckDuckGo, Bing Chatbots Keep Telling Stories About Lighthouse Keeper 'Elias Thorne'. We Might Know Why Scientists Just Accidentally Discovered a Strange, Hidden Rule of Human Nature Podcast: Google Employees Meme About How Bad Their AI Is Cops Keep Getting Arrested for Using Flock to Stalk People FCC Wants to Kill Burner Phones By Forcing Telecoms to Get All Customers’ IDs Judge Learns Lawyers on Both Sides of Case Used AI, Cancels Trial, Kicks Everyone Off the Case 'Sloppenheimer:' Amazon Employees Mock the Company’s AI on Slack This Company Will Add Phone, AirPod, and Smartwatch Trackers to License Plate Readers Microsoft Hacked to Deliver Malware to Claude and Gemini Users It’s So Not Over for Hollywood (with Devindra Hardawar) A Farmer Donated Land to Turn into a Park. The City Is Building a Massive Data Center Instead Scientists Discover Hidden Symmetry on Earth That Nobody Can Explain ICE’s Plan to Let Cops Around the Country Scan Faces to Verify Immigration Status Behind the Blog: Dangerous Memes The U.S. Military Quietly Turned GPS Into a Global ‘Numbers Station,’ Evidence Suggests K-pop Fans Are Calling Out Creepy Deepfakes of Idols Satya Nadella ‘Not Sure’ Who Said Microsoft Wanted to Make Addictive AI, Is Looking for Guy Who Did This The LEGO YouTube Scandal Has Broken Containment, Can No Longer Be Ignored Immigrant Rights Lawyers File Lawsuit Over Palantir’s ELITE Watch These Judges Rip Into Lawyers For Citing Cases That Don't Exist Google Employees Internally Share Memes About How Its AI Sucks Demand Is Booming for New No Tech, Repairable Tractor Podcast: Hackers Asked Meta AI To Let Them In. It Worked Companies Are Using Reddit to Manipulate ChatGPT and Google AI Search Google Is Quietly Buying Code From Play Store Developers to Train AI Microsoft Wants to 'Make People Addicted' to its New AI Assistant, Internal Documents Reveal Nvidia and Microsoft Researchers Say AI Agents Don't Care About Safety or Reliability Here is the Contract for Palantir’s Super API for the IRS Amazon Shuts Down Internal AI Leaderboard After Employees Cheated Hackers Simply Asked Meta AI to Give Them Access to High-Profile Instagram Accounts. It Worked We Sued ICE to Get Its Spyware Contract. The Agency Is Redacting Essentially Everything AI Grifters Are Making Anti-Data Center Slop With AI ‘Highly Plausible’ Aliens on Europa Are Earthlings’ Descendants, Study Says Behind the Blog: Being New and Some Numbers New Study Reveals the Manipulative ‘Dark Patterns’ of AI Chatbots Cities Are Covering Flock Cameras With Trash Bags The Sun Is Undergoing a Mysterious Change and Nobody Knows Why ‘Cracked Oura’ Is an App For Using the Oura Ring Without the Monthly Subscription Headway Therapy Patients Forced to Scan Their Faces to Keep Getting Care ‘Lobotomized’: Character.AI Is Showing What AI Enshittification Looks Like Podcast: How Deepfakes Destroyed a High School Millions of People Are Installing Malware on Their Partners’ Phones (with Zack Whittaker) ‘BusPatrol’ Put AI Cameras in Tens of Thousands of School Buses. Now They Want to Give Cops Access An Incomplete List of Successful Anti-Data Center Legislation ‘Corpse Point’ In the Arctic Is Melting, Disturbing Centuries-Old Bodies Here's the Bodycam Footage of the Cybertruck That Drove Into a Lake Behind the Blog: The Attention Wars How Deepfakes Tore a High School Apart
Companies Are Making Claude and Codex Talk Like Cavemen to Stop AI’s Soaring Costs
Joseph Cox · 2026-06-30 · via 404 Media

Companies are deliberately making their AI tools speak like cavemen in an attempt to stop burning through AI tokens and curb their massive expenditure on AI, 404 Media has found. The tool turns the usually verbose outpost of LLMs like Claude Code, Codex, or Gemini into a much more to the point answer. Think less “you’re right to push back, I was wrong,” and more “Hulk smash.”

Use of the caveman plugin is in direct response to the skyrocketing and unpredictable cost of AI. As 404 Media previously reported, companies are scrambling to stop spending so much on AI, with consulting giant Accenture finding much of the “soaring token spend” is thanks to people using AI to convert PDFs to presentations. People using caveman include developers at OpenAI, Nvidia, and GitHub, according to the tool’s creator. A senior OpenAI employee has even contributed code to the project, adding support for OpenAI’s Codex tool.

💡

Do you know anything else about token spend inside companies? I would love to hear from you. Using a non-work device, you can message me securely on Signal at joseph.404 or send me an email at joseph@404media.co.

“I made Caveman back in early April because I was using Claude Code heavily and noticed a lot of my token spend was going to unnecessary prose: pleasantries, hedging, transitions, and chatty language that does not really matter inside an agent loop,” Julius Brussee, the creator of caveman, told 404 Media. 

One company using caveman is electrical and digital infrastructure giant Legrand which, ironically, has entered the data center business. An internal Legrand memo shared with 404 Media tells employees “since the billing system changed and the new quotas were implemented, we all need to be mindful of our usage of AI so we don’t use up our entire budget allowance too quickly.” It goes on to list four things that will produce “high impact”: not always using the most powerful model; not always using high reasoning settings for the LLMs; using different more appropriate models for different tasks; and finally “use ‘caveman skill’ to reduce output consumption (without impacting code).”

In 404 Media’s tests of caveman with Claude Code, the plugin does make the LLM’s answers much more to the point. “Want changes to it?” the LLM asked after I told it to review some previously written code. “Uses official API, not scraping,” the LLM added, describing how the code worked. When I double checked caveman was installed Claude outputted, “Already active. What you need?”

Caveman can also display what it says is the total number of tokens saved. In my case, caveman said it had saved me around 5,800 tokens, or 65 percent. 

A screenshot of caveman in action.

“It makes the model speak less like a polite chatbot and more like a terse tool,” Brussee said. “Same substance, fewer words. In my evals, Caveman cut output tokens by roughly 65–75 percent versus default verbose output, and still beat a normal ‘be concise’ instruction. That number varies by workflow, but the effect was clear.”

Caveman users can pick their level of “grunt”: lite, full (which is the default setting), ultra, or Wenyan, which translates the output into classical Chinese characters (I verified this works and now have no idea what the Claude output says).

“The goal was to reduce output tokens without touching the parts where exactness matters: code, commands, paths, URLs, numbers, function names, and technical details. Caveman mostly compresses the surrounding language,” Brussee added.

Records on GitHub show that Shayne Sweeney, director of engineering at OpenAI, has contributed code to caveman. A commit a couple of months ago says, “Add Codex plugin support.”

A screenshot of caveman in action.

Caveman also offers a whole agent that condenses everything down to caveman language. “caveman-code shrink everything — full terminal coding agent, caveman top to bottom. ~2× fewer tokens than Codex on identical tasks. 20+ providers · plan mode · autopilot goal loop · MIT,” caveman’s GitHub repository says. Caveman can also be used with OpenClaw, the agentic AI tool that went massively viral earlier this year.

The plugin is obviously pretty funny but comes in response to a very real problem. In April, GitHub announced it was going to start charging customers per token rather than a flat subscription fee. Uber capped employee’s use of AI tools and the company’s CTO says Uber blew through its entire AI budget in just four months. Walmart also capped AI tool usage. And now you have companies using caveman.

“I’ve heard from many individual developers and engineers inside companies using or testing it, including people at OpenAI, NVIDIA, GitHub, and DEPT,” Brussee said.

In leaked audio obtained by 404 Media, Accenture positioned itself as the cure to this problem, even though it encouraged clients to adopt AI as quickly as possible in the first place. In that audio a senior employee said Accenture had a new opportunity with its clients “to really think about token economics.”

Last year OpenAI CEO Sam Altman suggested that people extending their own pleasantries to LLMs, like “please” and “thank you,” cost OpenAI tens of millions of dollars in electricity costs.

Legrand, OpenAI, Nvidia, and GitHub did not respond to requests for comment on their caveman use.

Caveman’s GitHub repository says near the end: “Caveman save you token, save you money.”

About the author

Joseph is an award-winning investigative journalist focused on generating impact. His work has triggered hundreds of millions of dollars worth of fines, shut down tech companies, and much more.

Joseph Cox