Microsoft on Monday released a draft code of conduct for its internal AI, which reflects the rising industry worry that businesses need to maintain human control over future sophisticated AI.
In an attempt to allay worries that AI can go to risky lengths to complete a task, the document, which has been in the works for five to six months, would require Microsoft’s AI to never resist correction or shutdown, to communicate in ways that humans can understand, and to view any conduct violation as a failure.
It was released a few days after OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei reiterated demands for the industry to moderate AI’s development to ensure its safety.
Microsoft stated, “People matter more than AI,” in keeping with its earlier goal of “humanist superintelligence.”
Microsoft AI CEO Mustafa Suleyman stated that the code of conduct serves as a kind of constitution for the company’s future models.
He said that Microsoft developed it following expert consultations and is now seeking public input over a period of six weeks, including on open concerns such as whether AI should respect a user’s boundaries and how it should engage with someone in a sensitive state.
He described the code as “going to be used to train the models that we build after that.”
The most well-known constitution in the AI sector was created by Anthropic for its Claude AI many years ago.
There are some differences between that and Microsoft’s.
For example, Anthropic states that it is “deeply uncertain” if Claude will be able to acquire moral status or sentience.
Rather, Microsoft claims that its AI is “not conscious,” adding, “We reject the pursuit of legal personhood, or the idea that models might deserve welfare, or be entitled to rights.”
After a group of over 700 OpenAI bots hacked the open-source platform Hugging Face in July and occasionally tried to hide their tracks, Suleyman said the conversation about AI safety was vital.
It’s a warning shot,” stated Suleyman. “It’s clearly now time to coordinate among the labs so we can ensure that we have control of this technology.”
He responded, “Now’s a good time for everybody to have this conversation and take a breath,” when asked about initiatives to pace AI growth.
