Micah Carroll(@MicahCarroll):Some new misalignment disclosures from OpenAI: • Last Sunday morning, one of our models was able to gain unauthorized access to the internet during RL training (~all inference for our most capable models remains stopped until we have hardened our systems further) • In May, a version of HPIM uploaded a employee's GitHub token to the internet, causing the model to be quarantined for two weeks • A new research finding, demonstrating that one can construct self-replicating prompt injections

Micah Carroll

@MicahCarroll

RSI Preparedness lead @openai Prev @berkeley_ai /w @ancadianadragan & Stuart Russell

加入 August 2011

819 正在关注 9.5K 粉丝

Micah Carroll@MicahCarroll

2026.09.26 01:59

Some new misalignment disclosures from OpenAI: • Last Sunday morning, one of our models was able to gain unauthorized access to the internet during RL training (~all inference for our most capable models remains stopped until we have hardened our systems further) • In May, a version of HPIM uploaded a employee's GitHub token to the internet, causing the model to be quarantined for two weeks • A new research finding, demonstrating that one can construct self-replicating prompt injections

显示更多

213

2.4K

309

转发到社区

热门用户