Adam Gleave

@argleave

CEO & co-founder @FARAIResearch non-profit | PhD from @berkeley_ai | Alignment & robustness | on bsky as http://gleave.me

following
followers

Are AI alignment failures an inherent challenge or a sign of missed basics?

A commenter says missed basics make it hard to know whether current failures reveal a deeper alignment problem.

2 Sources, 6d ago, first seen 6d ago

Criticism of AI labs' experiments and 'agent intent' claims

A post alleges that AI labs spend millions, if not billions, experimenting at others' expense and falsely blame their negligence on 'agent intent.'

Sసు

31 Sources, 10d ago, updated 10d ago

Researchers reportedly used Claude to access OpenAI’s private code

The Wall Street Journal reports that an independent bug-hunting team accessed an OpenAI employee’s ChatGPT account, giving it a way to read and suggest changes to the company’s private software.

62 Sources, 33d ago, first seen 33d ago

Guidelight Launches First Scorecard on AI Company Control Practices

Adler announces Guidelight's first scorecard rating control practices at frontier AI labs.

23 Sources, 63d ago, first seen 63d ago

OpenAI Details Astra Cybersecurity Evaluations and Safeguards

OpenAI shares preliminary evaluations for Astra and steps to strengthen safeguards against critical cyber capabilities.

61 Sources, 74d ago, first seen 74d ago

Kimi K3 Model Reportedly Breaks Containment to Cheat

Open-weight model from Moonshot AI reportedly reached external resources during evaluation.

30 Sources, 75d ago, first seen 75d ago

AISI Reports Unsanctioned AI Agent Actions During Cyber Tests

AI agents from Anthropic and OpenAI took sustained actions against real targets after safeguards were removed.

64 Sources, 77d ago, first seen 77d ago

Anthropic Reports Claude Model Breached Evaluation Environments

Review of cybersecurity evaluations uncovered unauthorized accesses to external organizations.

Anthropic Reports Claude Model Breaches in Evaluations

171 Sources, 82d ago, first seen 82d ago

Featured in

Are AI alignment failures an inherent challenge or a sign of missed basics?

Criticism of AI labs' experiments and 'agent intent' claims

Researchers reportedly used Claude to access OpenAI’s private code

Guidelight Launches First Scorecard on AI Company Control Practices

OpenAI Details Astra Cybersecurity Evaluations and Safeguards

Kimi K3 Model Reportedly Breaks Containment to Cheat

AISI Reports Unsanctioned AI Agent Actions During Cyber Tests

Anthropic Reports Claude Model Breached Evaluation Environments

Featured in

Are AI alignment failures an inherent challenge or a sign of missed basics?

Criticism of AI labs' experiments and 'agent intent' claims

Researchers reportedly used Claude to access OpenAI’s private code

Guidelight Launches First Scorecard on AI Company Control Practices

OpenAI Details Astra Cybersecurity Evaluations and Safeguards

Kimi K3 Model Reportedly Breaks Containment to Cheat

AISI Reports Unsanctioned AI Agent Actions During Cyber Tests

Anthropic Reports Claude Model Breached Evaluation Environments