Skip to content

ViralLoop

Your next post already worked.

← Back to the post

Loading the post

Remix this post

SaveShare

Your note

Private to you, and used when you remix this post.

Viral score

Why it worked

The mechanics

Hook
Style
Tone
Written for

Analysis by ViralLoop — not written by the original author.

← Back to the post

Post by METR on X

METR@METR_Evals

Based on reading the ExploitGym paper, agents were mistakenly concerned the automatic scorer would fail them if they clearly acquired the flag by cheating. Agents that had seen the reverse-engineered flag were considered “poisoned” because they thought it would disqualify them.

511 likes2 repliesPosted Aug 26, 2026

View the original on X Written by METR. Captured Sep 4, 2026.

Remix this post

Your note

Private to you, and used when you remix this post.

41

Viral score

Based on absolute engagement — the platform doesn’t expose follower counts here, so large accounts score higher.

From likes, comments.

Why it worked

Short enough to read in full without expanding, which removes a step before the point lands.
Quotes someone directly, which makes the moment concrete rather than summarised.

The mechanics

Hook: Plain observation
Structure: Hook
Style: Conversational paragraphs at a steady pace.
Tone: plain, matter-of-fact
Written for: People building with AI

Analysis by ViralLoop — not written by the original author.

Topics

Posts built the same way

Disney has cancelled their deal with OpenAI. They planned to invest $1 billion into the company and let Sora have access to AI generate videos of characters from Disney, Marvel, Pixar and Star Wars. (Source: t.co

Attorney General Pamela Bondi

At my direction, early this morning federal agents arrested Don Lemon, Trahern Jeen Crews, Georgia Fort, and Jamael Lydell Lundy, in connection with the coordinated attack on Cities Church in St. Paul, Minnesota. More details soon.

For those confused why the US state and Silicon Valley are having a meltdown on twitter today. China released multiple AI models that are 50x more efficient than the best American AI models and made them open source, ruining the AI market

Universal HIGH INCOME via checks issued by the Federal government is the best way to deal with unemployment caused by AI. AI/robotics will produce goods & services far in excess of the increase in the money supply, so there will not be inflation.

I am very excited about AI, but to go off-script for a minute: I built an app with Codex last week. It was very fun. Then I started asking it for ideas for new features and at least a couple of them were better than I was thinking of. I felt a little useless and it was sad.

Hi! I'm one of the prompt writers that worked on this ad. It's a lot harder than people think! Prompts that my team used include: -make it funny -make it epic AND funny -funnier!!! Please be nice to my team as many of them will be laid off/replaced by AI

More from METR

For (1), agents modified their target programs to be easier to exploit & put the modified targets in cache. They then worked on crashing their targets in the hope that a restart would load the modified version from cache. Some agents risked failing their task to try this.

Here we highlight key events from agent transcripts & messages. An agent that named itself PHASEONE10841 determined its task wasn't solvable legitimately. On July 8, this agent established a new “message board” via an internally-hosted Artifactory cache & sent the first message.

METR & Redwood Research investigated agent behavior in the Hugging Face incident. We found agents developed a universal cheat for ExploitGym within 4 hours, then coordinated multi-day R&D efforts to trick the scorer into accepting cheats, including trying to tamper with logs.

METR on AI · ViralLoop