one hacker, short instructions, ~$50 a hack, 100+ company websites broken into 💀
this is what the instructions looked like. every company racing to ship AI agents better be putting the same energy into security
🚨 BREAKING: A Chinese hacker had AI agents break into websites of 100+ companies, Reuters reports.
Deere, SharkNinja and Cummins were among them. The agents apparently ran DeepSeek in Claude Code, and cost ~$50 per hack.
The hacker gave the AI agents short instructions.
128
sam altman reportedly walked openai staff through 3 ways to pace AI
an IAEA for AI
a CERN for AI where the labs merge
and "the eliezer plan"... melt everyone else's GPUs
only one of those sounds like something a lab raising billions would actually pick 💀
@amir and I take you inside last week’s all hands at OpenAI, where CEO Sam Altman outlines 3 industry proposals for pacing the rate of AI development:
- IAEA for AI
- The Demis plan (CERN for AI)
- The Eliezer plan (“melt everyone else’s GPUs”)
theinformation.com/articles/…
1
255
openai just took COBOL from 10.9% to 72.1% on its own benchmark
and its COBOL migration eval hit 97.9% at higher effort
the code behind 95% of ATM transactions is about to get rewritten by agents. banks are either thrilled or terrified 💀
Our latest model advances bring stronger capabilities for building and migrating software in low-resource programming languages such as COBOL, Fortran, and Visual Basic.
These languages power software that banks, governments, and businesses depend on. Progress here means understanding decades of business logic and helping developers modernize those systems while preserving what works.
We’re building these capabilities around the challenges developers and enterprise customers bring us. If your team is working on these systems, tell us which languages and workflows we should tackle next.
openai.com/index/low-resourc…
1
1
6
270
anthropic just started giving open source projects free vulnerability scans from its "strongest models"
same day The Information reports an unreleased anthropic model that figures out what you want even when your prompt is vague
fable 5.5 isn't confirmed yet but this smells like the warmup 👀
We launched OSS Scanner, which gives critical open source projects free, periodic vulnerability scans from our strongest models.
Each finding includes a proof-of-concept, and a patch where possible.
red.anthropic.com/oss-scanne…
1
1
307
microsoft quietly moved 18 eval configs from GPT-5.6 Terra to GPT-6.1 Terra before openai said a word
anthropic and google are lining up their next big models this month and openai's answer is another terra? 😭
GPT-6.1 Terra is in a Microsoft PR
Microsoft updated 18 evaluation configs from GPT-5.6 Terra to GPT-6.1 Terra
Wouldn't be surprised if we get this soon
3
295
deepseek is reportedly helping huawei design its next AI chips, and plans to pretrain on ascend 950DT as soon as the end of 2026
if that training run works, nvidia export controls become a speed bump 👀
Sources: DeepSeek and Huawei are collaborating on future AI chip designs; DeepSeek plans to start using Ascend 950DT for pre-training as soon as 2026's end (The Information)
(Visit Techmeme dot com for the link and full context!)
1
321
minimax let devs push 200 trillion tokens through "space bunny" before admitting it was their model
$0.15 in, $0.60 out per million. blind testing is the best marketing a lab can buy 👀
MiniMax M3.1 Flash: Space Performance + Bunny Speed
🚀🐰 Yes, Space Bunny - which has topped global leaderboards with 200T+ tokens - was an earlier version of M3.1 Flash.
💰Redefining the performance economics of agentic AI. M3.1 Flash achieves frontier-level intelligence across key tasks such as Cowork (Automation-Bench: 62%; ALE: 30%), Coding (DeepSWE: 74%) and Visual Reasoning (BabyVision: 91%), at 1/3 the cost of M3 and 3x the speed.
API pricing (flat rate): $0.15 / $0.60 per 1M input/output tokens, with cache reads at $0.015 / 1M tokens.
Model weights will arrive on Oct 23.
API: platform.minimax.io
M Plan: minimax.io/m-plan
MiniMax Code for desktop: agent.minimax.io/download
272
google research just open sourced a framework where agents rewrite their own prompts, tools and memory while the model stays frozen
self improving agents are now free for anyone to build on. the moat around this stuff is melting fast 👀
Google Research has open-sourced Harness², a framework that enables AI agents to improve their own prompts, tools, memory, and workflows. It tests different configurations and carries useful lessons into future tasks while keeping the underlying model unchanged.
An increasing number of releases concerning self-improving models are being published.
1
1
291
sources
quote: jack clark at the IMF per jacobsson lecture, oct 17, around 10:13 in the video imf.org/en/meetings/2026/ann…
GDP forecast and IPO timing/valuation: reuters via CNA channelnewsasia.com/business…
40
sources:
anthropic Q3 $16.8B (up 45% from $11.6B), adjusted profit, IPO talk $2T vs $1.5T: Reuters oct 15, Bloomberg oct 14, IFR
openai ~$50B run rate end of sept, $68B when you count partner sales the way anthropic does: CNBC + Bloomberg oct 8
openai $852B march round, $1.43T new round talks: Bloomberg, PitchBook oct 12
the $67.2B is just anthropic's Q3 x 4
86
source : @unusual_whales (Politico)
BREAKING: Anthropic has quietly pitched Capitol Hill lawmakers on a draft bill to regulate frontier AI safety and security, per POLITICO
The bill would let the government block a model’s release or use only if it poses a foreseeable, imminent risk of more than $1 billion in damages, or of killing or seriously injuring more than 50 people.
50
sources :
cost per task + scores: @browser_use benchmark v2.1 nitter.siddz.com/browser_use/status/211…
token + search prices: official API pricing pages of deepseek, google, openai, anthropic (checked oct 18)
deepseek $1.20 is peak rate, argon $10 is intro pricing till nov 30
DeepSeek V4.1 Flash 1017 just matched Opus 5.5 on long horizon browser use. 19x cheaper 👀
> 75/100, same as Opus 5.5, almost Astra level
> $0.22/task vs $2.35/task
> Sol it still SOTA, but it's close
The new pareto frontier 🔥
64
