The wmdp benchmark: Measuring and reducing malicious use with unlearning
The White House Executive Order on Artificial Intelligence highlights the risks of large
language models (LLMs) empowering malicious actors in developing biological, cyber, and …
language models (LLMs) empowering malicious actors in developing biological, cyber, and …
[PDF][PDF] Managing ai risks in an era of rapid progress
In this short consensus paper, we outline risks from upcoming, advanced AI systems. We
examine large-scale social harms and malicious uses, as well as an irreversible loss of …
examine large-scale social harms and malicious uses, as well as an irreversible loss of …
Refusal in language models is mediated by a single direction
Conversational large language models are fine-tuned for both instruction-following and
safety, resulting in models that obey benign requests but refuse harmful ones. While this …
safety, resulting in models that obey benign requests but refuse harmful ones. While this …
Managing extreme AI risks amid rapid progress
Artificial intelligence (AI) is progressing rapidly, and companies are shifting their focus to
developing generalist AI systems that can autonomously act and pursue goals. Increases in …
developing generalist AI systems that can autonomously act and pursue goals. Increases in …
Protecting society from AI misuse: when are restrictions on capabilities warranted?
M Anderljung, J Hazell, M von Knebel - AI & SOCIETY, 2024 - Springer
Artificial intelligence (AI) systems will increasingly be used to cause harm as they grow more
capable. In fact, AI systems are already starting to help automate fraudulent activities, violate …
capable. In fact, AI systems are already starting to help automate fraudulent activities, violate …
International Scientific Report on the Safety of Advanced AI (Interim Report)
Y Bengio, S Mindermann, D Privitera… - arXiv preprint arXiv …, 2024 - arxiv.org
This is the interim publication of the first International Scientific Report on the Safety of
Advanced AI. The report synthesises the scientific understanding of general-purpose AI--AI …
Advanced AI. The report synthesises the scientific understanding of general-purpose AI--AI …
Lora fine-tuning efficiently undoes safety training in llama 2-chat 70b
AI developers often apply safety alignment procedures to prevent the misuse of their AI
systems. For example, before Meta released Llama 2-Chat, a collection of instruction fine …
systems. For example, before Meta released Llama 2-Chat, a collection of instruction fine …
Connecting the dots: Llms can infer and verbalize latent structure from disparate training data
One way to address safety risks from large language models (LLMs) is to censor dangerous
knowledge from their training data. While this removes the explicit information, implicit …
knowledge from their training data. While this removes the explicit information, implicit …
Acceptable Use Policies for Foundation Models
K Klyman - Proceedings of the AAAI/ACM Conference on AI, Ethics …, 2024 - ojs.aaai.org
As foundation models have accumulated hundreds of millions of users, developers have
begun to take steps to prevent harmful types of uses. One salient intervention that foundation …
begun to take steps to prevent harmful types of uses. One salient intervention that foundation …
On the limitations of compute thresholds as a governance strategy
S Hooker - arXiv preprint arXiv:2407.05694, 2024 - arxiv.org
At face value, this essay is about understanding a fairly esoteric governance tool called
compute thresholds. However, in order to grapple with whether these thresholds will achieve …
compute thresholds. However, in order to grapple with whether these thresholds will achieve …