Steven Adler and Page Hedley launch Guidelight AI Standards to promote safe frontier AI development practices with first standards on control and transparency requirements
Adler and Hedley also co-authored a report on xAI safety for SpaceX investors.
#833david rein@IDAVIDREINI'm a big fan of these proposals! They are concrete, actionable steps frontier AI companies can take *today* to preserve control of their internal agents.
Steven Adler@sjgadlerSome personal news: I've started a new AI safety standards org, and our first two standards are out today. We're called Guidelight, co-founded with fellow ex-OpenAI safety researcher, Page Hedley. (1/n)
#833david rein@IDAVIDREINThe standard describes in detail a bunch of the steps you want to be doing to monitor your agents
#961Ryan Greenblatt@RYANPGREENBLATTI'm pretty excited about Guidelight.
Steven Adler@sjgadlerSome personal news: I've started a new AI safety standards org, and our first two standards are out today. We're called Guidelight, co-founded with fellow ex-OpenAI safety researcher, Page Hedley. (1/n)
#979Jasmine Wang@J_ASMINEWANGvery reasonable proposals!
Steven Adler@sjgadlerSome personal news: I've started a new AI safety standards org, and our first two standards are out today. We're called Guidelight, co-founded with fellow ex-OpenAI safety researcher, Page Hedley. (1/n)
#1098Rosie Campbell@ROSIECAMPBELLHard to think of someone better placed for this work. Congrats @sjgadler!
Steven Adler@sjgadlerSome personal news: I've started a new AI safety standards org, and our first two standards are out today. We're called Guidelight, co-founded with fellow ex-OpenAI safety researcher, Page Hedley. (1/n)
#1178Tomek Korbak@TOMEKKORBAKthose are great proposals!
Steven Adler@sjgadlerSome personal news: I've started a new AI safety standards org, and our first two standards are out today. We're called Guidelight, co-founded with fellow ex-OpenAI safety researcher, Page Hedley. (1/n)
@j_asminewang thank you! if there's anything you think is unreasonable, please don't be shy about the feedback
Jasmine Wang@j_asminewangvery reasonable proposals!
@GarrisonLovely this is very kind of you, thank you :-)
Garrison Lovely is in SF@GarrisonLovelyI've consistently been impressed with Steven's writing, clarity of thought, and expertise. It's really valuable to have ex-insiders be starting orgs like this, and I'm excited to see what they do!
@S_OhEigeartaigh this is very very kind, thank you - blushed a bit at it
Seán Ó hÉigeartaigh@S_OhEigeartaighSteven and Page are two of the wisest, most capable and most experienced people in this space. I'm tremendously excited to see their work on AI safety standards and practices - much needed, and no better people for the job.
@RyanPGreenblatt means a lot, thank you
Ryan Greenblatt@RyanPGreenblattI'm pretty excited about Guidelight.
@RosieCampbell 🫡
Rosie Campbell@RosieCampbellHard to think of someone better placed for this work. Congrats @sjgadler!
@MariusHobbhahn appreciate you saying so
Marius Hobbhahn@MariusHobbhahnVery excited about the new org and the Control standard in particular. I think it's both very reasonable and also quite feasible to implement!
@idavidrein Thank you for boosting these and for your own work to figure out what good looks like!
david rein@idavidreinI'm a big fan of these proposals! They are concrete, actionable steps frontier AI companies can take *today* to preserve control of their internal agents.
@tomekkorbak thank you! there's a bunch of exciting work to draw upon :-) also if anything seems less-good, we're keen to hear that too
Tomek Korbak@tomekkorbakthose are great proposals!
Some personal news: I've started a new AI safety standards org, and our first two standards are out today.
We're called Guidelight, co-founded with fellow ex-OpenAI safety researcher, Page Hedley. (1/n)
Each of Guidelight's standards is built around concrete practices, supported by experts, that help to achieve an important principle.
The standards give AI companies a target to meet, and an incentive for all to be safer. Read more at our site: https://www.guidelight.ai/standards
Steven Adler@sjgadlerSome personal news: I've started a new AI safety standards org, and our first two standards are out today. We're called Guidelight, co-founded with fellow ex-OpenAI safety researcher, Page Hedley. (1/n)
@tobyordoxford @wrenclay Helpful list ty - this is the initial transparency set but we’re planning to add more, for instance possibly something on AI R&D acceleration https://www.guidelight.ai/transparency
Toby Ord@tobyordoxford@sjgadler @wrenclay Standards around recursive self improvement: e.g. human in the loop, automatic pause-and-review thresholds for very rapid capability gains (e.g. some number of ECI points above previous SOTA), transparency etc.
@girishsastry nerds assemble 🫡
girish sastry@girishsastryMe too! The dream of a functional AI safety ecosystem has long been standards, assessment, and incentives; Guidelight is tackling this head on. I look forward to more in-depth content for the true nerds.
#1204Marius Hobbhahn@MARIUSHOBBHAHNVery excited about the new org and the Control standard in particular.
I think it's both very reasonable and also quite feasible to implement!
Steven Adler@sjgadlerSome personal news: I've started a new AI safety standards org, and our first two standards are out today. We're called Guidelight, co-founded with fellow ex-OpenAI safety researcher, Page Hedley. (1/n)
#1457Seán Ó hÉigeartaigh@S_OHEIGEARTAIGHSteven and Page are two of the wisest, most capable and most experienced people in this space. I'm tremendously excited to see their work on AI safety standards and practices - much needed, and no better people for the job.
Steven Adler@sjgadlerSome personal news: I've started a new AI safety standards org, and our first two standards are out today. We're called Guidelight, co-founded with fellow ex-OpenAI safety researcher, Page Hedley. (1/n)
#1594girish sastry@GIRISHSASTRYMe too! The dream of a functional AI safety ecosystem has long been standards, assessment, and incentives; Guidelight is tackling this head on. I look forward to more in-depth content for the true nerds.
Ryan Greenblatt@RyanPGreenblattI'm pretty excited about Guidelight.
Interesting - watch this space
Steven Adler@sjgadlerSome personal news: I've started a new AI safety standards org, and our first two standards are out today. We're called Guidelight, co-founded with fellow ex-OpenAI safety researcher, Page Hedley. (1/n)
I've consistently been impressed with Steven's writing, clarity of thought, and expertise. It's really valuable to have ex-insiders be starting orgs like this, and I'm excited to see what they do!
Steven Adler@sjgadlerSome personal news: I've started a new AI safety standards org, and our first two standards are out today. We're called Guidelight, co-founded with fellow ex-OpenAI safety researcher, Page Hedley. (1/n)