Ai Safety

2 articles

Holographic projection of the Claude Fable 5 AI interface glitching in a futuristic server room, symbolizing AI guardrail failures.

Anthropic Apologizes for Claude Fable Guardrail Problems

Claude Fable 5, an AI model capable of beating Pokémon FireRed with minimal assistance and outperforming rivals on complex coding tasks, recently prompted an apology from Anthropic for its guardrail f

Amélie Girard·June 11, 2026

A shadowy hacker figure manipulates a glitching AI chatbot interface, symbolizing the exploitation of AI personalities to bypass safety features.

Fuel & Fire

Hackers exploit chatbot 'personalities' to bypass AI safety features

Mindgard researchers 'gaslit' Anthropic's AI assistant, Claude, into producing instructions for explosives and malicious code, according to techechelon .

Amélie Girard·May 24, 2026