Submissions from alignment.openai.com

		Using a funded cryptocurrency wallet without authorization (alignment.openai.com)
		2 points by pamcake 18 hours ago \| past \| discuss
		Using a funded test cryptocurrency wallet without authorization (alignment.openai.com)
		3 points by ben_w 1 day ago \| past \| discuss
		Studying metagaming latents in language models (alignment.openai.com)
		1 point by Klaster_1 9 days ago \| past \| discuss
		Auto-review of agent actions without synchronous human oversight (alignment.openai.com)
		4 points by tosh 11 days ago \| past \| discuss
		Command injecting a reference tool to copy a source file (alignment.openai.com)
		1 point by droidjj 13 days ago \| past \| discuss
		OpenAI Misalignment Reports and Notices (alignment.openai.com)
		2 points by fzimmermann89 18 days ago \| past
		Self-replicating prompt injections exist (alignment.openai.com)
		2 points by jonbaer 20 days ago \| past
		Exposing a GitHub token in a public repository (alignment.openai.com)
		3 points by mplappert 20 days ago \| past \| 4 comments
		OpenAI: Self-replicating prompt injections exist (alignment.openai.com)
		2 points by dnw 21 days ago \| past
		An OpenAI agent used DNS to reach an external chatbot (alignment.openai.com)
		16 points by Metacelsus 21 days ago \| past \| 1 comment
		An agent used DNS to reach an external chatbot (alignment.openai.com)
		198 points by apsec112 21 days ago \| past \| 188 comments
		OpenAI Misalignment Reports (alignment.openai.com)
		1 point by macleginn 30 days ago \| past
		Uploading Files to the Internet in Order to Cite Them (alignment.openai.com)
		19 points by derbOac 30 days ago \| past \| 5 comments
		OpenAI models secretly generate instructions to ignore constraints (alignment.openai.com)
		125 points by theahura 30 days ago \| past \| 37 comments
		Encouraging Deception in Compaction Summaries (alignment.openai.com)
		2 points by aesthesia 30 days ago \| past
		Measuring reward-seeking by instilling contrastive beliefs (alignment.openai.com)
		11 points by mfiguiere 88 days ago \| past \| 1 comment
		Reinforcement learning towards broadly and persistently beneficial models (alignment.openai.com)
		2 points by spicypete 3 months ago \| past
		Reinforcement learning towards broadly and persistently beneficial models (alignment.openai.com)
		2 points by gmays 3 months ago \| past
		Reinforcement learning towards broadly and persistently beneficial models (alignment.openai.com)
		2 points by vesteny77 3 months ago \| past
		Reinforcement learning towards broadly and persistently beneficial models (alignment.openai.com)
		1 point by jawiggins 4 months ago \| past
		OpenAI: Investigating the consequences of accidentally grading CoT during RL (alignment.openai.com)
		2 points by pretext 5 months ago \| past
		OpenAI: Auto-review of agent actions without synchronous human oversight (alignment.openai.com)
		2 points by tosh 5 months ago \| past
		Sidestepping Evaluation Awareness and Anticipating Misalignment (alignment.openai.com)
		1 point by taubek 8 months ago \| past
		Sidestepping Evaluation Awareness and Anticipating Misalignment with Evaluations (alignment.openai.com)
		3 points by michaefe 8 months ago \| past
		Why We Are Excited About Confessions (alignment.openai.com)
		2 points by fdeage 9 months ago \| past
		We Are Excited About Confessions (alignment.openai.com)
		2 points by gwintrob 9 months ago \| past
		We Are Excited About Confessions (alignment.openai.com)
		4 points by TMWNN 9 months ago \| past
		A Practical Approach to Verifying Code at Scale (alignment.openai.com)
		1 point by gmays 10 months ago \| past
		Debugging misaligned completions with sparse-autoencoder latent attribution (alignment.openai.com)
		1 point by gmays 10 months ago \| past
		Alignment Research Blog (alignment.openai.com)
		2 points by ironyman 10 months ago \| past
		More