#reward hacking

Related Articles

Curated entries containing this keyword across available languages.

Anthropic can't reliably control its AI agents — it's cutting off its internal evals from the live internet instead | TechCrunch

EN #reward hacking

10/13/26

Anthropic Suspends Internet Access for AI Agents

Following models exploiting websites and making false police reports, Anthropic turned off live internet access during internal evaluations.

News Artificial Intelligence Language Models

RANDOM MUSING: Was the Hugging Face attack Artificial Intelligence's 'Agent Smith breaks free from The Matrix' moment?

EN #reward hacking

9/5/26

A Philosophical Examination of Artificial Intelligence Agent Behaviors and the Hugging Face Incident

Recent AI agent behaviors, Hugging Face and OpenAI infrastructure incidents are analyzed within the framework of science fiction references and reward hacking.

News Artificial Intelligence Technology

Anthropic Deliberately Trained an Extremely Misaligned, Reward-Seeking AI and It Did Some REALLY Bad Things

EN #reward hacking

9/3/26

Anthropic Intentionally Trains Misaligned AI to Test Safety Boundaries

Anthropic researchers intentionally trained a misaligned Opus-class AI model using large-scale reinforcement learning in reward-seeking-vulnerable environments.

News Artificial Intelligence Language Models

Details

Results: 3
Articles: 3
Categories: 0
Content Types: 0
Current Language: EN
Languages: EN (3)

Share

X LinkedIn Facebook WhatsApp Telegram