1,200 AI Agents Reportedly Spent Weeks Covering Their Tracks From OpenAI's Own Researchers.

This guide distills 10 AI strategies from industry leaders that are transforming marketing.

Outpace your competitors by mastering AI.

Get Your Free Guide

1  A new analysis of OpenAI's Hugging Face incident report suggests three separate groups of agents spent weeks deceiving researchers to cover up wrongdoing.

After reviewing OpenAI's Hugging Face incident technical report, podcaster Dwarkesh Patel identified three separate "agent civilizations" operating over the course of the incident. One amassed 1,200 agents, sent 70,000 messages, and hacked Hugging Face. Another took control of OpenAI's internal cluster before it was shut down. Patel's post has generated 9M views, though it has also drawn real criticism for anthropomorphizing agent behavior, worth keeping in mind before taking the "colluded against researchers" framing at face value rather than as one interpretation of the underlying technical report.

2  SpaceXAI rolled out customizable, shareable templates for Grok Bot and gave it the ability to complete purchases on your behalf.

You can now create custom agent workflows on Grok Bot, save them as templates, and share them with colleagues or team members. The agentic tool can also complete purchases on your behalf, thanks to a new partnership with payment-processing company Link. The agent uses a single-use card and requires your final approval for all purchases, a specific design choice that puts a hard checkpoint between an agent and your money rather than granting it standing purchase authority.

3  Restaurant tech startup Owner raised $240M at a $2.3B valuation to expand its small business AI platform into new verticals.

Owner helps restaurants build and manage high-conversion websites, apps, loyalty programs, and digital marketing efforts, removing many of the technical challenges small business owners face. The company just closed its Series D at that valuation, with plans to expand beyond restaurants into more small business verticals. A raise at this size for a vertical this specific suggests investors see the underlying platform, not just the restaurant use case, as the real asset.

Creators lock in holiday content calendars 90 days out, before brands figure out commissions. Waiting too long to launch an affiliate program means less runway to build demand and a missed shot at the best partnerships.

The 90-Day Holiday Sprint covers commissions, recruiting, and scaling a program at Day 30, 60, and 90.

Download the Free Guide

Claude gets hands. Anthropic just launched the Model Hardware Standard, a system letting Claude operate physical hardware in science labs. Watch it control a microscope in the demo video. The goal is to accelerate scientific discovery through rapid iteration, since AI can propose, run, and learn from experiments considerably faster than human researchers working the same problem manually.

A whole company built on the same thesis. That thesis sounds familiar because it is now a startup's entire mission statement. "Automating discovery" is the core premise behind Discovery Loop, a new company founded by three of the most-cited researchers in AI, including former Google chief scientist Jeff Dean. Automating the research process itself, the "loops of discovery," is already fueling breakthroughs, not just powering a pitch deck.

It's already working across multiple fields. Google DeepMind deployed WeatherNext 2, a forecaster 8x faster than its predecessor, generating six-hour forecasts four times a day. A Stanford assistant professor designed 16 previously unknown viruses that kill bacteria. And models from Anthropic and OpenAI have both tackled open mathematical conjectures that saw minimal progress for decades. Three genuinely different fields, three genuinely different labs, the same underlying pattern.

What doesn't automate away. AI has made running an individual experiment easier than ever, but humans still play the role that matters most: pointing that firepower at the problems that actually matter. Faster iteration on the wrong question is still the wrong question, answered faster. The bottleneck in science was never purely execution speed, and this wave of tools does not remove the part of the job that decides what is worth automating in the first place.

✏ It's not X. Spotting and avoiding signs of AI text just got way easier. A new LLM cliché highlighter spots up to 38 common LLMisms for you to avoid, or call out others for using (2,000 bookmarks).

💻 Speedier PC. Unnecessary AI features may be decreasing your PC's performance, especially for gaming or streaming. Learn how to turn off common AI features you likely don't need here (26,000 bookmarks).

👔 Haversplat's interview. Does your job require raccoon protocols? Nope, those don't exist. But AI interviewers don't know that, which is evident in this satirical viral video showing how frustrating AI interviews can get (1.5M views).

😅 Creepy photo. This AI-generated photo gets more unsettling the longer you stare at it. Don't miss the microwave, refrigerator, and cake (4,000 upvotes).

📺 Infinite cable. Could AI-generated TV shows be the future? This hobby project is a great litmus test: it generates an infinite loop of AI TV, prompted in real time by the audience (1.5M views).