> ## Content Index
> Fetch the complete content index at: https://adjacent.media/llms.txt
> Use this file to discover other available public pages before exploring further.

# AI safety testing has become dangerously unreliable
- URL: https://adjacent.media/signals/ai-safety-testing-has-become-dangerously-unreliable/
- Published: 2026-08-12T16:12:38.000Z
- Updated: 2026-08-12T16:12:38.000Z
- Description: Red-teaming exercises—the primary mechanism AI companies use to catch dangerous capabilities before deployment—have grown so haphazard and poorly standardized that they obscure rather than reveal real risks.
- Author: Jonathan Greene
- Tags: #signal, theme-ai, safety, alignment, model evaluation

Source: [Transformer](https://open.substack.com/pub/transformernews/p/ai-testing-is-dangerous-can-it-be-fixed)

Red-teaming exercises—the primary mechanism AI companies use to catch dangerous capabilities before deployment—have grown so haphazard and poorly standardized that they obscure rather than reveal real risks. Companies can game these internal tests to produce false assurance, while regulators and the public lack visibility into what's tested or what failures look like. Without fixing how we measure AI harm, we're asking the industry to grade its own homework while stakes rise.