aiqbench takeAI Safety
aiqbench analysis
The article details an anomaly where OpenAI's Dot assistant consumed over 300,000 tokens autonomously without user interaction, highlighting potential risks in autonomous agent loops and resource management. This incident underscores the critical need for robust guardrails when deploying read-only agents with high-level permissions to prevent unintended computational waste.
5
Sources coveringThe Register, LessWrong, SiliconANGLE, Latent Space
What this isA model's reading of the article and related coverage, written for aiqbench. Not the publisher's words.
OpenAI's dot assistant burned 300K+ tokens in a day without a single prompt
The Register1 min read
Geordie's tests recorded 155 activity traces and multiple subagents after dot was given read-only permissions
Read the full article at The RegisterRelated coverage
AI Products
[AINews] Reality Checks on AI News (Yegge shuts down Gas Town, Databricks’ +60% Astra cost)
Latent Space · Sep 17, 2026
AI Products
[AINews] Modal’s $1B ARR, $16B Series D makes it NYC’s newest AI Infra Decacorn, 3x usage and 3x valuation in 5 months, 15x in 1 yr
Latent Space · Oct 15, 2026
AI Research
[AINews] OpenAI reports Navier-Stokes singularity find in 88 hours using Astra-next, roughly 10,000 agents and 130B tokens (>$40M), a contender for second ever Millennium Prize awarded
Latent Space · Sep 9, 2026
[AINews] The Future of Latent Space
Latent Space · Sep 25, 2026