The 'Gay Jailbreak' Technique Exposed: New AI Safety Vulnerability Draws Industry Attention
In 2025, an LLM attack technique dubbed the 'Gay Jailbreak' has sparked widespread discussion. Attackers exploit AI mode…
📂 LLM News
Latest articles in LLM News
IBM Releases Granite 4.1 Series Open-Source Models
IBM has officially launched the Granite 4.1 series of large language models, spanning multiple parameter scales and rele…
'Gay Jailbreak' Exposes Deep Contradictions in LLM Safety Alignment
A large language model bypass technique dubbed the 'Gay Jailbreak' has sparked heated debate in the AI community. The me…
GPT-5.5 Matches Mythos Preview in Cybersecurity Threat Assessment
The latest cybersecurity tests reveal that OpenAI's GPT-5.5 has reached a level comparable to the much-hyped Mythos Prev…
Amazon Nova Models Introduce LLM-as-a-Judge for Reinforced Fine-Tuning
Amazon dives deep into the RLAIF technical approach, leveraging LLMs as judges to perform reinforced fine-tuning on its …
Google Releases Gemma 4 Open-Source Model Series
Google DeepMind has officially launched the Gemma 4 series of open-source models, touting them as the most capable open …
UK AI Security Institute Evaluates GPT-5.5 Cybersecurity Capabilities
The UK AI Security Institute has completed its cybersecurity capability assessment of OpenAI's GPT-5.5, finding its vuln…
OpenAI Tried to Make ChatGPT a Nerd — It Became Obsessed with Goblins Instead
OpenAI recently attempted to inject a "nerdy" personality into ChatGPT, but the experiment backfired when the model deve…
Why Was ChatGPT Obsessed With 'Goblins'? OpenAI Has Fixed the Issue
Recently, a large number of users discovered that ChatGPT was frequently and inappropriately mentioning "goblins" in its…
OpenAI Fixes Bizarre ChatGPT 'Goblin' Bug
OpenAI recently discovered a peculiar bug in ChatGPT where the model frequently and inappropriately referenced "goblins"…
IBM Launches Granite 4.1: 8 Billion Parameter Model Rivals 32B MoE Models
IBM has officially released the Granite 4.1 series, in which the 8B dense model matches or even surpasses the performanc…
OpenAI Set to Launch GPT-5.5-Cyber, a Cybersecurity-Focused AI Model
OpenAI founder Sam Altman has announced that GPT-5.5-Cyber will be rolled out to key cyber defenders in the coming days.…