CRITICALAi
Global

Perturbation Probing: A New Diagnostic for the Fragility of LLM Safety

·Source: Unit 42 (Palo Alto)

Updated:

Executive Summary

New research reveals that AI safety refusal lives in a thin neural layer, highlighting the critical need for external, multi-layered security. The post Perturbation Probing: A New Diagnostic for the Fragility of LLM Safety appeared first on Unit 42 .

Analysis

New research reveals that AI safety refusal lives in a thin neural layer, highlighting the critical need for external, multi-layered security. The post Perturbation Probing: A New Diagnostic for the Fragility of LLM Safety appeared first on Unit 42 .
Source Attribution

Originally published by Unit 42 (Palo Alto) on Aug 28, 2026.

Related Threats

LOWAi

Anthropic warns infostealer malware is hijacking Claude sessions to drain usage

Anthropic is warning some Claude users that infostealer malware on their PCs has stolen active Claude login sessions, allowing attackers to access accounts and consume their usage. [...]

BleepingComputer
MEDIUMAi

Anthropic is cutting Claude Code's current weekly limits by 17%

Anthropic is permanently increasing Claude Code's standard weekly usage limits by 25% for Pro, Max, Team, and seat-based Enterprise plans, but it's not as good as it sounds. [...]

BleepingComputer
CRITICALAi

ServiceNow patches three maximum severity flaws that could put enterprise data at risk

Code injection and SQL injection attacks have been around for decades, and they are still tried-and-true ways for attackers to compromise systems. ServiceNow’s latest trio of maximum severity flaws shows that even AI-era platforms remain vulnerable to these techniques: The software provider has released patches for three bugs in its ServiceNow AI Platform that could be exploited via low-complexity

CVE-2026-18885CVE-2026-18886
CSO Online