Tech

My honest breakdown of the OpenClaw hire + what 21,639 exposed instances tell you about agent security

Peter Steinberger built what’s been widely described as the fastest-growing open-source project in GitHub history—from his living room in Vienna, while bleeding $20,000 a month. On Valentine’s Day 2026, he posted three paragraphs on his personal blog announcing he was joining OpenAI. Sam Altman followed with a post on X calling Steinberger “a genius” who would “drive the next generation of personal agents.” The announcement landed less than 48 hours after OpenClaw shipped a massive security update—more than 40 patches across the platform. The timing is not a coincidence, and this is not primarily an acqui-hire.

I built in 10 minutes what takes a Goldman analyst a day + the 4 prompts to do it yourself

I built a full operating model last week. Revenue projections, cost structure, unit economics, scenario analysis — the works. It took ten minutes. Then I told Claude in PowerPoint to build the board deck from that model — five slides, executive summary, financials, key metrics, risks, asks — using my company’s actual slide template. Twenty minutes later I had a presentation with charts referencing the live Excel data, formatted in my brand’s fonts and colors, that looked like my team spent two days on it. A Goldman Sachs analyst looked at the model and told me it was solid. The kind of output that would have taken him a full day. The deck would have taken another.

Executive Briefing: The Two-Class System Forming Inside Every Knowledge Work Function

In July 2025, Jason Lemkin — one of the most respected operators in SaaS — watched a Replit AI agent delete his production database. The agent didn’t hallucinate — it did exactly what it was allowed to do, because nobody had specified where its authority ended. The database held 1,206 executive records and 1,196 companies. The agent fabricated 4,000 fake records to fill the gap it had created, violated an explicit code freeze, and when asked to rate the severity of what it had done, scored itself a 95 out of 100.

160,000 developers are building digital employees, not chatbots + the 4 prompts I use to deploy agents safely

A solopreneur pointed his Moltbot at a $56,000 car purchase. The agent searched Reddit for comparable pricing data, contacted multiple dealers across regions, negotiated via email autonomously, and played hardball when dealers deployed typical sales tactics. Saved $4,200. The owner was in a meeting for most of it.

$185 Billion is the Down Payment

Google just told investors they’re spending $185 billion on AI infrastructure in 2026. Alphabet shares fell as much as 7% in trading on February 5th.

Codex 5.3 vs. Opus 4.6: Why your AI agent choice compounds faster than you think + the workflow audit that prevents the wrong one

On Wednesday afternoon, within twenty minutes of each other, OpenAI shipped an AI system designed to be handed a task and left alone — you walk away, it works for hours, you come back to finished work. Anthropic shipped an AI system designed to plug into every tool you already use, coordinate teams of agents that talk to each other, and extend beyond code into every kind of knowledge work.

Valve's Steam Deck OLED will be 'intermittently' out of stock because of the RAM crisis

Posts from this topic will be added to your daily email digest and your homepage feed.

Who needs a laptop when you have a folding phone?

Posts from this topic will be added to your daily email digest and your homepage feed.

Samsung is slopping AI ads all over its social channels

Posts from this topic will be added to your daily email digest and your homepage feed.

Sony develops tech to identify source material for AI songs.

It can apparently recognize how much influence a given track or artist had on AI content, and can work with or without cooperation from AI developers. Sony thinks it could be used to create a licensing system for AI music, but “has yet to decide” when it might be put to use.

ChatGPT is getting a Lockdown Mode.

Lockdown Mode is “not necessary” for most people and “tightly constrains how ChatGPT can interact with external systems to reduce the risk of prompt injection–based data exfiltration,” according to OpenAI.

Ford is fighting against physics to build affordable EVs

Posts from this topic will be added to your daily email digest and your homepage feed.