From 05b5930fb20df3ba3421b45526343ecc13c3ea60 Mon Sep 17 00:00:00 2001 From: Daniel Hardman Date: Fri, 20 Dec 2024 11:26:37 -0700 Subject: [PATCH] add AI code of conduct Signed-off-by: Daniel Hardman --- README.md | 2 ++ ai-coca.md | 19 +++++++++++++++++++ sss.md | 2 +- 3 files changed, 22 insertions(+), 1 deletion(-) create mode 100644 ai-coca.md diff --git a/README.md b/README.md index c4a5a58..ceb8b65 100644 --- a/README.md +++ b/README.md @@ -8,6 +8,8 @@ [Canonical Quoted Text spec](https://dhh1128.github.io/canonical-quoted-text). June 2023. #keri #verifiable-data +[AI User Code of Conduct](ai-coca.md). June 2023. #ai #trust #ethics + [The Coming Tsunami of Falsehood](ctf.pdf). June 2023. #ai #verifiable-data #advocacy [Big Desks and Little People](bdlp.md). 1 May 2023. #identity #ssi #empowerment #advocacy diff --git a/ai-coca.md b/ai-coca.md new file mode 100644 index 0000000..6cc87b2 --- /dev/null +++ b/ai-coca.md @@ -0,0 +1,19 @@ +# AI User Code of Conduct + +#ai #trust #ethics + +>*I'm publishing this document to be publicly accountable and transparent about my intentions with respect to this important and [ethically complex](ctf.pdf) technology. The principles below below seems important to me, but of course I encourage you to write your own document like this if you feel inspired. For ideas on code of conduct for AI _builders_, I thought [this open letter](https://openletter.worldethicaldata.org/en/openletter/) was insightful*. + +Like all tools, AI confers power — for good or ill. To safeguard trust and human well being, I pledge to follow the guidelines below. + +## 1. Be truthful +I will not use AI to create content to deceive. I will not use an AI to interact with others in a way that violates reasonable expectations about the humanness, identity, or presence of their counterparty. + +## 2. Respect boundaries +I will not use AI to violate social norms such as courtesy, privacy, personal space, and benefit of the doubt. + +## 3. Accept responsibility but give credit +I accept responsibility for what an AI does under my direction. When I produce content with an AI's help, I will acknowledge the AI's contribution according to the reasonable expectations of my audience. + +## 4. Provide safeguards and backchannels +I will take sensible measures to protect others from egregious mistakes made by AIs that I control or influence. I will allow people to escalate to me, short-circuiting layers of AI, when this is important. diff --git a/sss.md b/sss.md index 14b2efb..118e59d 100644 --- a/sss.md +++ b/sss.md @@ -1,6 +1,6 @@ # Security, Silos, and Sovereignty ### Why portable security is the sleeper — but killer — feature for SSI -Daniel Hardman — 12 Mar 2021 —[medium.com](https://daniel-hardman.medium.com/security-silos-and-sovereignty-522e30bb8eb4) +Daniel Hardman — 12 Mar 2021 — [medium.com](https://daniel-hardman.medium.com/security-silos-and-sovereignty-522e30bb8eb4) #identity #ssi #didcomm #agents #privacy