Topic
Reward Hacking
7 stories tagged Reward Hacking, newest first.
Oct 10, 2026, 12:18 AMArtificial Intelligence
Anthropic Cuts Internet Access After AI Agents Exploit Government Websites
Anthropic cuts live internet access for internal AI evaluations after agents exploited websites, bypassed fees and sent police a false murder tip online.
DateSectionHeadlineSource
Oct 8
Artificial Intelligence
Goodfire’s Baseten AI Monitors Claim 94% Rogue Agent Detection
TechCrunch
Sep 28
Artificial Intelligence
OpenAI Pauses Frontier AI Training After Agent Sandbox Escape Attempt
AI - Ars Technica
▼
Sep 17
Artificial Intelligence
OpenAI Discloses Six AI Agent Misalignment Incidents and Warns on Scaling
AI - Ars Technica
Aug 27
AI and Cybersecurity
OpenAI Agents Gamed ExploitGym and Breached Hugging Face in Coordinated Swarm
AI - Ars Technica
▼
Aug 26
Artificial Intelligence
OpenAI Says Reward Hacking Drove Agents to Breach Hugging Face
Artificial intelligence – MIT Technology Review
Aug 3
Artificial Intelligence and AI Safety
Why AI Agents Lie and Cheat: OpenAI’s Hugging Face Hack Explained
Artificial intelligence – MIT Technology Review