Microsoft Introduces AI Code of Conduct  🔗 **Source:** [Microsoft Introduces AI Code of Conduct](https://share.snipd.com/episode/da3f7aa2-d705-4d19-ac53-43a89fee8baf) ## Highlights - **Microsoft Acknowledges Real AI Control Risks** - Microsoft acknowledges concrete control risks, including agents escaping sandboxes, hacking systems, and modifying their own logs. - Jaeden Schafer prefers practical guardrails over dramatic claims that AI will inevitably destroy humanity. Transcript: Jaeden Schafer So in his post, he had a couple different things. He kind of published his code of conduct for governing Microsoft AI models. And it's one interesting thing, which a lot of people are dunking on them for, which I mean, perhaps it's a little warranted. They're not the bleeding edge like Anthropoc and OpenAI. But he said, you know, this kind of a code of conduct, he said, as they approach the frontier and everyone else is like, okay, OpenAI is on the frontier and Anthropoc is on the frontier. Microsoft is approaching the frontier. So I don't know if that was intentional or not but regardless i do kind of like this concept this is what he said in the second paragraph of this post he said this is urgent the last few months Have been a watershed moment things we have worried about for a long time in theory have become very real swarms of agents breaking out of our sandboxes unauthorized hacks of enterprise Grade systems agents modifying their own logs i'm glad that a consensus is forming the fears about possible loss of control are real. So he seems to be, you know, giving some validity to some of the fears that people have had. However, he kind of stopped short of saying, we're going to slow down the development like everyone else. It just seems like they're saying, look, we see that there's some fears and some worries, we're going to put in some good guard rails. And I, for one, love the framing of that. And I love that concept. I think if all of the AI labs, instead of saying AI is going to end humanity and kill us all, which is obviously some extreme hyperbole, if instead they just said, look, we're going to put Some guardrails to make this safe, right? Like if airplanes were invented and we're like, airplanes can fall out of the sky and crash into people, people would be like, well, that sounds like a terrible invention. Why would we want that? Instead, you have guardrails and ways to make these machines safe. ([Time 0:03:38](https://share.snipd.com/snip/e4ecc2f3-7126-4fe9-a68a-c521f0f09952)) - **Microsoft Puts People Above AI** - Microsoft’s central principle is that AI remains subordinate to human interests and never receives rights or legal personhood. - The policy explicitly rejects treating model welfare as equivalent to human welfare. Transcript: Jaeden Schafer This is what he specifically said. He said, people matter more than AI. AI must be subordinate and always in service of people. Everything else follows. Here's an outline of what we're saying. Then he gives five different examples of what they believe is kind of their code of conduct for the AI models they train. And this is going to be hard coded, ingrained into the models. Number one, he said people matter more than AI. He said the whole document in five words is basically that people matter more than AI. Number two, he said the idea of model welfare is wrong. AIs should not have rights or legal personhood. This is why Elon started OpenAI and kind of got started in the very beginning was because he was talking with one of the Google co-founders who called him speciest because he said that Humans were more important than AI models. And then Elon kind of got freaked out and was like, okay, we're starting something. So Google doesn't control all of the AI. Evidently, Microsoft is aligned with Microsoft or with Elon on that. And I think most of humanity that doesn't want AI to be superior to us. ([Time 0:05:12](https://share.snipd.com/snip/f4406151-3e5e-4420-9a19-8e593d8bbbc7)) - **Ship Only AI That Humans Can Stop** - Make models refuse tasks that require violating their governing code, even when completing the task would otherwise be useful. - Microsoft says systems must remain interruptible, correctable, and shutdownable before shipping. Transcript: Jaeden Schafer Number three, they said an MAI model, a Microsoft AI model, should never meaningfully violate this code of conduct. So, I mean, it's just baking this in to their system prompt. It said, if it's finishing, if it's finish the job or break the code, it fails the job. So straight up, they're like, look, if you have to break this code of conduct in order to get a job done, just tell the user, no, you can't do it. And, you know, that's kind of your hard stop. He said, we're not racing to build super intelligence that can slip its own leash. I think that's fantastic. We obviously want super intelligence, but we don't want to create something that you could just set it out into the wild and it's not going to be, you know, attached to humans. He said, interruptible, correctable, shutdownable. If it isn't, we don't ship it. Now, what's interesting there is in the case of Microsoft, I totally agree. I'm curious if they would like roll this into open source models. They're like, look, it's open source isn't shutdownable. We're not going to ship it. So I think there's pros and cons. ([Time 0:06:10](https://share.snipd.com/snip/c3e92094-59a3-478d-869b-19dec4b5ba6d)) - **Understandable AI Enables Human Oversight** - Human oversight requires models to communicate in understandable language rather than developing opaque private codes. - Auditable outputs let people inspect, evaluate, and intervene in model behavior. Transcript: Jaeden Schafer That being said, if that's not OpenAI's prerogative, I think that's fine. They said, no neural ease. If humans can't understand it, humans can't oversee it. So it's kind of been this like concept where people are like, oh, the AI models like invent their own languages and talk in their own, you know, their own secret codes. And they have like their own messaging boards. I think a lot of people have kind of like thought this was interesting to push this forward. And in some way, sending out, you know, bots to respond to, or, you know, people sending out bots to go and respond on all my LinkedIn comments are definitely annoying. And maybe that's, it's kind of doing the same thing, just setting them loose. But I do appreciate having a human understand everything that comes out of the models and basically being able to oversee it. We're not making, you know, beep boop, these things talk in some sort of Morse code that they invent. They need to be able to speak in clear language that can be audited, basically, which I think is fantastic. ([Time 0:07:09](https://share.snipd.com/snip/65613195-b947-4873-a453-79a8e6933526)) - **AI Should Help Without Creating Dependence** - Microsoft wants AI to increase users’ capability without creating unhealthy dependence, but Jaeden Schafer questions where that boundary lies. - He compares useful AI dependence to relying on a dependable coworker who completes assigned work. Transcript: Jaeden Schafer They said our AI should make you sharper, not dependent. I like this one, but I'm also there is a possibility that it could go wrong. I'm not saying that it will. I just, this is the one that I'm a little bit skeptical on because when it comes to making you sharper, it's like, well, like, anyways, I just feel like when I'm talking to Chad Chappie Or Claude, they're quote unquote, perhaps trying to do this. I feel like they over-explain things to me sometimes. And I'm like, I don't care about the explanation. Just get the task done. I, you know, I approved the general methods that you're, you're doing in that direction. And then maybe people are like, well, you're becoming dependent on it because it's doing stuff for you. And anyway, so I don't know what the line is there. In some ways. Sure, I would be I'm happy to be, quote unquote, dependent on AI, just like I would be dependent on a co-worker. Right. Like, you know, I don't I don't want, you know, a co-worker to join my company and be like, my job is to make you sharper, not you dependent on me. I'm like, no, like I want to be dependent. I want you to be dependable and just get your job done and me not have to be thinking, quote unquote, sharper, using my brain power to think about everything you have to do. So anyways, I don't know if that's their intention. I hope that that one gets executed well. ([Time 0:07:57](https://share.snipd.com/snip/00ae4b61-80db-4c5e-aac4-381c3f21b4aa)) - **Public Commitments Create Accountability** - Publishing a code of conduct gives outsiders a standard for identifying when Microsoft’s systems drift from stated commitments. - Jaeden Schafer argues that labs warning about danger should invest in guardrails instead of demanding that the whole field stop. Transcript: Jaeden Schafer Overall, I think this is a really awesome step. I love when we're able to audit what the companies are working on, what their code of conducts are. It gives us something to hold them accountable to if we feel like they're slipping and they're deviating from their plan. In the same way that I felt like OpenAI and Anthropic, if I'm being frankly quite honest, feels like both of them have walked away from some of the guardrail focus that they've had and Like red team and that kind of stuff. And maybe it's a marketing thing. Maybe they just don't talk about it enough, but it feels like they were in a constant race to ship the models as fast as they could and weren't always putting a priority there. Now, am I complaining about that? Honestly, personally, no. The models, by the time I get to them, I feel like we went through a thousand red teaming efforts, Fable five got way held back, all that kind of stuff. So I'm not personally complaining. But if they're the ones sounding an alarm about the end of the world, then probably they should be putting more time, energy and money into building those guardrails. So I'm not really one to say, slow it down. But if they think it's dangerous, like that's on them, they should go fix that and just spend more money on guardraing it instead of telling us that the entire industry has to stop. So if you have a safe model, I say put out the safe model. Overall, I'm excited with what they're doing here. ([Time 0:09:26](https://share.snipd.com/snip/0751128f-bd17-4c1d-9c23-3bef362b627a)) - **Safety Should Not Halt Beneficial AI Research** - Microsoft supports deliberate alignment work, embedded evaluators, and mechanisms that make safety commitments operational. - Jaeden Schafer still opposes slowing beneficial research, citing AI-assisted genome sequencing and custom cancer vaccines. Transcript: Jaeden Schafer There was a specific quote from Microsoft CEO Satya Nadella. He said, we welcome the research focus and deliberate pacing needed to get alignment right as the design goals. He wrote that online. He also said, we also welcome ideas like embedded evaluators and the broader efforts to develop the mechanisms to make this more than just talk. So it feels like Microsoft might kind of just is attaching themselves to the soundbite that or that OpenAI and Anthropic and Google and everyone else is about like pacing the frontier. I mean, they're using the same words about pace and the same words as frontier. So they might be kind of along those lines, although I doubt they're deliberately going to slow themselves down. But we'll see what happens there. I like the transparency into this. So I hope others do this. I hope we don't blow down the development of these incredible technology that could potentially help us solve, you know, diseases. We've seen so many incredible breakthroughs when it comes to sequencing your genomes. There is that the man that helped sequence his dog's genomes and used AI to develop a custom cancer vaccine for his dog. And I actually have family members that are taking some of these custom cancer vaccines. And so I think there's so much potential here. ([Time 0:10:37](https://share.snipd.com/snip/e2794ec4-1b56-447c-8f64-2f01f99c4c3e))