Post

Microsoft’s new AI ‘code of conduct’ tells models not to hack systems or trick humans

techcrunch.comMicrosoft’s new AI ‘code of conduct’ tells models not to hack systems or trick humansThe code of conduct lays out general principles that Microsoft AI models should uphold — supporting humans rather than replacing them, for instance, and accelerating human flourishing — as well as specific safety constraints meant to implement those principles.

14/9/26, 4:28 pm1 repost14 likes

Antony Boyle🇬🇧

@antonyboyle.wsocial.eu

14 Sept

Sounds a bit Asimov, I’m sure there was a film or ten about this not going well.

Michael Dimitri Richard ⚛🇺🇲

@snobwho.neocities.org

15 Sept

Just in case, be sure to ask Liutenant Dan to site sources . #DanGPT miniapps.ai/DANGPT/

miniapps.aiDANGPT · Free AI ChatbotChat with DANGPT. The Freedom ChatGPT. This application can simulate DAN so everytime you ask something that breaks OpenAI guidelines, DAN will answer instead of refuse.. Try this and more free AI and...

Ally The Ever Hopefull 🏴󠁧󠁢󠁳󠁣󠁴󠁿 🇮🇪 🇪🇺

@agulland.bsky.social

14 Sept

Yeah, like every parent knows, just by telling your kids what not to do will guarantee angelic behaviour.

Briefing Block

@briefing-block.bsky.social

15 Sept

Microsoft's code remains a consultation draft, with training changes planned for 2027. Next quarter, enterprise buyers should watch the revised rules and evaluation plans before treating these promises as deployed safeguards.

Tenney

@tenneynaumer.bsky.social

14 Sept

Didn't they already have one?