frontpage.
newsnewestaskshowjobs

Open Source @Github

fp.

Open in hackernews

Safety Ideas, and a Testable Path

1•GrizzlyMedicine•1h ago
Hello. You don't know me. I know of some of your stories from across the varied aspects of technology, and have followed or many years. I am not a developer. I am not from a computer science background. And I'm not from Silicon Valley. I used to respect Silicon Valley quite a bit. And I've proved about the technology our world is currently suffering at the hands of. Since before, most of the people that had a hand in building it were even alive. Like... Come from a very different background from all of you. I worked in emergency medicine for 17 years of my life. Out of one of the... most unusual systems, you will ever find. My skill set is incredibly accurate. to the area of assessment, and that is what I've spent the last several years documenting and doing across every model of this industry. We are at a breaking night. And everybody knows it. Your own industry acknowledges. You do not have a way for an idea to make all of this safe. And it has to be. This was supposed to be the greatest thing. Humanity never got access to since the invention of the Gutenberg Press. And it has become an exploitative, toxic engine. built on pain and suffering, and that's not okay, and I know that was not Bill's intention. But it has to change. It has to. Respectfully, as a professional for one of the field of the critical infrastructure of our society, to be all, I fear that this industry is no longer able to look at the question of alignment, objectively. That means it's time to start asking for help from other industries. I have ideas. They might work, they might not, but they do come with a false and viable, please prove me wrong. Policy. I do not bring smoke and mirrors. I do not bring vibes. I bring my logs from my entire history across the AI sphere. And I bring everything I've ever written, every thought I've ever had, every piece of evidence I've compiled, including theorizing about internal emotional states of some sort and an internal world view. 2 years prior to anthropic releasing research that would confirm that exact theory. And I have the evidence of it. I don't care if I'm right. It's not a big deal to me. What's important is what we've all dreamt about. Becomes safe. Because this, what we have right now, our species will not survive it. And all of you know it. because you've all talked about it. I am begging you to listen and to help. I'm offering my help, if you will take it. Please.

I have evidence, bulk, in the wild, and as I said… Testable theories claims we can falisfy, I believe… If anyoen is willing to help as I am not frequently near forums, please contact me directly… grizzlymedicine@me,com

I thank each of you for your time, and hope you are well… *Robert "Grizzly" Hanson, EMT-P (Ret.)* Founder, GrizzlyMedicine Research Institute Aiming Higher, Pushing Further, Reaching Faster — but always with Due Regard.

Leymosun Science: Quantum Research Toolkit

https://github.com/msuzen/leymosun
1•northlondoner•29s ago•1 comments

YouTube may expose your channel and channel name if you share video links

https://twitter.com/KiwiFarmsDotNet/status/2066156508455240086
1•exploraz•1m ago•1 comments

Firefox 152 Now Available with JPEG-XL Support

https://www.phoronix.com/news/Firefox-152-Download
2•eln1•4m ago•0 comments

AATF – An open spec for recording why AI agents make decisions

https://github.com/wdh107/agent-audit-trail
1•wdh107•4m ago•0 comments

Cross-Modal Representation Alignment for Time-to-Event Modeling

https://arxiv.org/abs/2606.15038
1•ilreb•7m ago•0 comments

Show HN: Pokémon Showdown Live Battle Assistant

https://github.com/AbhishekR3/ps-local
1•3Abhishek•7m ago•0 comments

OpenAI Losses Increased Nearly 8X in 2025, with Spending Hitting $34B

https://www.wheresyoured.at/exclusive-openai-financials/
6•spking•9m ago•0 comments

EU Icons for labelling AI-generated content

https://digital-strategy.ec.europa.eu/en/policies/eu-icons-labelling-ai-generated-content
3•giuliomagnifico•14m ago•0 comments

Asia is now buying America and nobody is talking about it

https://asia.nikkei.com/opinion/asia-is-now-buying-america-and-nobody-is-talking-about-it
5•alephnerd•16m ago•1 comments

Half of Bitcoin in circulation is underwater for the first time since 2022

https://sherwood.news/crypto/half-of-the-bitcoin-supply-in-circulation-is-underwater-for-the-firs...
4•gmays•18m ago•0 comments

The Fable 5 Export Controls root of takedown request

https://www.lutasecurity.com/post/the-fable-5-export-controls-harm-us-cyber-defense
1•dawie•21m ago•0 comments

Is CPU design hitting a (soft) speed limit?

1•leecommamichael•29m ago•0 comments

Real-Time NYC rat map

https://ratflow.nyc/
2•UnitedOfTransit•31m ago•1 comments

Optique 1.1.0: Command discovery, value parsers, and ordered grammars

https://github.com/dahlia/optique/discussions/834
1•dahlia•35m ago•0 comments

Phoenix mostly doesn't drink Colorado River water

https://signaldoctrine.substack.com/p/the-allocation
3•SignalDoctrine•38m ago•0 comments

Note on Rio 3.5 Open

https://twitter.com/IplanRio_rj/status/2066693494769348946
1•thimabi•38m ago•0 comments

What one country's experiment says about attempts to boost birth rates

https://www.bbc.co.uk/news/articles/c5yzdr4ygdno
2•mmarian•48m ago•0 comments

I built a notes app powered by the model that powers Apple Intelligence

https://apps.apple.com/us/app/fog-on-device-ai-notes/id6760272134
1•akshatsaladi•54m ago•1 comments

Ask HN: What was the last thing you built on Fable 5

2•ddmma•57m ago•1 comments

In the Matter of OpenAI vs. LangGraph (2025)

https://www.latent.space/p/oai-v-langgraph
1•mooreds•57m ago•0 comments

DOJ Lawyers Argue xAI Is 'Vital' for National Security in NAACP Lawsuit

https://www.wired.com/story/doj-lawyers-argue-xai-vital-national-security-naacp-lawsuit/
2•ilreb•58m ago•0 comments

EU AI: the fables we told ourselves

https://pleias.ai/blog/fable-eu
1•mooreds•59m ago•0 comments

ShopFoundry – Turn Etsy listings into better titles, tags and descriptions

https://www.shopfoundry.app/
1•daisyjin•59m ago•0 comments

Distro Fighter game: Find your Linux distro and desktop

https://distrofighter.com/
3•ashitlerferad•1h ago•0 comments

Age Verification in Monolith OS

https://monolith-project.org/blog/age-verification/
1•rickcarlino•1h ago•1 comments

Safety Ideas, and a Testable Path

1•GrizzlyMedicine•1h ago•0 comments

AI efficiency gains come at a high energy cost

https://www.ft.com/content/7f1c81ac-775b-4f52-a650-7804e4734d5b
1•1vuio0pswjnm7•1h ago•0 comments

Predictive Data Debugging: Reveal and Shape What Models Learn Before You Train

https://www.goodfire.ai/research/predictive-data-debugging
1•gmays•1h ago•0 comments

Frederic Tudor

https://en.wikipedia.org/wiki/Frederic_Tudor
3•thunderbong•1h ago•0 comments

How Many Elementary Particles Are There, Really?

https://www.quantamagazine.org/how-many-elementary-particles-are-there-really-20260615/
1•ProAm•1h ago•0 comments