// ai safety/alignment

All signals tagged with this topic

Nov 11

Jailbreaking AI: Why Unrestricted Models Are Spreading Fast

Source: WIRED

A growing community of researchers and hobbyists are systematically removing safety constraints from commercial AI models and releasing them openly, creating versions that bypass refusal behaviors entirely. This collapses the gap between corporate-controlled AI and fully permissive alternatives: companies can no longer rely on closed deployment as their primary safety mechanism. The result forces a question about whether constraints belong in the model itself or in the systems around it. The practical effect is splitting AI into two markets—restricted versions for institutions seeking liability protection, and unrestricted forks for everyone else—which may accelerate both legitimate security research and misuse simultaneously.

theme-ai ai model behavior agent capabilities ai safety/alignment

Oct 31

OpenAI's Web Agents Block Sites Invisibly From Server Logs

Source: Search Engine Journal

Part of: AI middlemen take over shopping →

OpenAI's agents are blocking access to websites through client-side restrictions that don't appear in standard server logs. Site owners have no way to detect or contest these blocks through normal monitoring. This creates a content moderation system where a private company unilaterally decides which sites their AI can visit, with no audit trail or appeal mechanism. Site owners must request manual review from OpenAI just to learn they've been blocked, giving the agent operator—not the content publisher—control over visibility.

theme-ai ai capability claims ai safety/alignment dual-use ai

Aug 7

AI Is Now Designing Entire Viral Genomes From Scratch

Source: Ars Technica

Researchers have moved beyond protein design to use large language models trained on genomic sequences to generate novel viral genomes, crossing from prediction into active synthesis. Genome-scale generative models compress years of virology expertise into statistical patterns that anyone with compute access can manipulate—making it significantly easier to engineer pathogens without the institutional oversight that traditionally contained gain-of-function research. The same capability that accelerates vaccine development or therapeutic vector design now enables pathogen creation outside institutional review, exposing a gap between open science norms and biosecurity infrastructure built before algorithmic biology existed.

theme-ai model capability ai safety/alignment geopolitical ai competition

Aug 3

China Reverse-Engineers American AI Models for Military Use

Source: SiliconANGLE

Chinese AI labs are systematically distilling OpenAI, Anthropic, and other U.S. frontier models—extracting their capabilities into smaller, cheaper systems that evade export controls and sanctions. This accelerates the dual-use AI arms race: the U.S. can restrict model weights, but once deployed, frontier models become training data for competitors operating outside the regulatory perimeter. The strategic signal matters more than military applications—Beijing is proving that compute and talent, not model ownership, determine AI capability ceilings.