I think Dario is writing with much of these things in mind, but using this as a specific framed opportunity to enable a better interim and long-term outcome for this planet with us still on it. When the internal motives for Dario are fully shared on the other side of the hill, I think we will find his writing to be more about influencing and shaping policy to steer the world in the best way possible within his ability, control, and knowledge. It’s a commonly held believe with the effective altruism community, which Anthropic team was formed under, to use the policy levers to positively influence the world. If I read this latest writing from Dario through that lens, there are deeper concerns with the AI tooling that go well beyond what’s noted in this singular post and are likely the motivations for re-attempting a slowdown or pacing. I too generally agree with for many reasons like allowing the young OpenAI engineers to make mistakes like the Hugging Face incident in July and put better processes in place (ie where is your decent multi-level fencing controls and harness best practices when disabling security features and where is your energy/spend limits for achieving the mission objective in scope and not warming the world for no reason!) and more time to engineer and deploy better hardware to not warm the world instead of burning through fossil fuels since now data centers in the US can isolate pollution from the grid to exempt themselves from federal regulation.
That was sunset and replaced by Antigravity. FWIW until I abandoned it knowing the sunsetting, I was able to get good behavior out of Gemini CLI with overriding the system prompt. The default prompt crippled the harness with very poor instructions, but there was a hidden ENV to override it. Replacing it with Claude Code like prompts based on the model selected, it ran at a much higher intelligence level full stack with significantly less errors.
It may be a comforting idea to believe that our affection should require no assistance at all, and expect from all others that it should flow quite effortlessly from a caring mind. But human capacity, I have come to realize with both age, health, and genetic variations is rarely so simple a matter. When a person admits quite vulnerably that their memory falls short as the author has, yet their desire to care for their relationships remains entirely intact, there is something very moving in that alone. But to spend ten years quietly building as the author has is an act of love for people they care for are not lost to memory and forgotten in time.
“Meta internally projected late last year that it would earn about 10% of its overall annual revenue – or $16 billion – from running advertising for scams and banned goods, internal company documents show.”
Having grown a large healthcare review platform, I can attest to the success we had mapping specific policy violations to natural language is incredibly useful. At scale, patients having terrible situations and/days can write about in ways that can be deeply unhealthy for the community or the doctors reading/receiving the feedback and sometimes very threatening beyond that purposes for the community. We built a custom ML engine to handle our levels of traffic for reviews, which was among the largest in the US typical ranking top 3 on Google for the domain keywords. Back when BERT was the edge, a policy-adaptive model like this one from Mistral would have been an incredible cold-start solution. Most sites never have the massive volume nor budget needed nor skillset needed before you can train domain-specific models that outperform OOTB solutions. Generally, most people and site mean well and try to do well, so empowering those people with models like this can help the collective in my opinion, so I’m happy to see this released in this manner myself.
A bit of editorial and cultural note from a US native, the subsection of the original article “Teach discrimination, not memorization” is better worded as something like 'Differentiation' or 'Distinction' instead of ‘Discrimination’. In English, the word 'discrimination' can (and in this social context may) imply social prejudice or unfair treatment. I think this may have been a bit of carry over from the rather benign French translation of “Enseigner la discrimination" which I also see awkwardly translated in the paper as well.
Discriminative has a meaning in machine learning that I think is relevant here. There are "generative" models like LLMs that are learning joint probabilities P(X, Y) and "discriminative" models like logistic regression that learn conditional probabilities P(Y | X)
Thanks for the clarification. I’m very happily wrong here, and I appreciate the correction. This does emphasize from an editorial review that the double meaning can be distracting for someone who isn’t very deep in the terminology of this particular area, so it may still be worth rewording the subheading for a broader audience which will be interested in this model.
I don’t think Cupertino will settle for stock. I think they’ll demand cash and an agreement that OpenAI abandon or reboot their hardware project. In the meantime, Apple gets an open kimono into everything OpenAI has planned.
This could actually be the fuckup that kills OpenAI as an independent company. The threat of a cash judgement gums up not only an IPO, but also debt-based fundraising. (We equity guys are idiots, so we’ll probably keep writing cheques until the market turns.)
This could go way beyond any potential hardware project, depending on whats actually true from the allegations and how much of Apple's trade secrets have been used or shared within OpenAI.
There were rumors a while back of OpenAI building some integrations into macOS akin to the Shortcuts app, and who knows what other computer use type projects. They could very well have been using information from within Apple for that.
It's kind of a fruit of a poisonous tree problem. If OpenAI used any of Apple's secrets broadly, Apple could ask the court for an injunction to block the deployment of any software from OpenAI. If discovery proves the IP contamination spread into other areas within OpenAI, it could completely freeze all of their deployments.
If Apple actually has the receipts here they could realistically bring down OpenAI entirely.
If OpenAI used any of Apple's secrets broadly, Apple could ask the court for an injunction to block the deployment of any software from OpenAI. If discovery proves the IP contamination spread into other areas within OpenAI, it could completely freeze all of their deployments.
I find this so funny. Can you share an example or two of where something like this has ever happened to one of the largest companies in the world? There is no universe where some judge orders that OpenAI can no longer deploy any software, for anyone (including huge swaths of the federal government) because of the alleged actions of a few people, and it’s allowed to stand. Zero chance.
I’ll say one thing for Sama: he might have a lot of haters, but it’s not that hard to prove them wrong with predictions like these.
> Can you share an example or two of where something like this has ever happened to one of the largest companies in the world?
Waymo (Google) vs. Uber & Anthony Levandowski. Google was granted an injunction that ordered Uber to immediately halt all development on self driving tech.
Apple also bankrupted Pystar for stealing trade secrets.
Levandowski would have served actual jail time if he didn't get a pardon.
Property rights and trade secret laws don't just magically vanish because a company has government contracts or a high valuation.
That’s not true! Only Levandowski was barred from working on self-driving, the judge specifically elected NOT to force Uber to stop all self-driving work [1], let alone stop all software work, let alone this being anything that was a mortal threat to Uber. Uber was also worth about $50 billion at the time, about 1/20th the size of OpenAI.
The point isn’t that an injunction is impossible, but the idea that Apple will get an injunction that shuts down all software deployments at OpenAI, leading to the end of the company, is pure fantasy.
My prediction: this will result in a relatively modest undisclosed settlement and OpenAI won’t abandon, or even modify, their hardware product because of this. And Apple definitely won’t get an “open kimono” to everything OpenAI has planned.
This is two of the largest, most powerful, most well-capitalized companies on the planet in a legal fight. People are taking sides because of their hatred of Sama, and it results in these bizarro takes that OpenAI is finished as a company lol.
> This is two of the largest, most powerful, most well-capitalized companies on the planet in a legal fight.
OpenAI being well-capitalized made me laugh out loud, thanks for the humor :)
Apple has a ridiculous amount of cash, AA credit, and makes $100B of net income per year. OpenAI has none of those attributes, what they do have is investors willing to bankroll them, for now.
OpenAI just closed the largest funding round in SV history, $122 billion. The idea that you’d find it laugh out loud funny to call them well-capitalized just reveals your bias.
OpenAI has plenty of cash to fight this. Apple having hundreds of billions doesn’t actually mean anything in a legal fight that might cost OpenAI nine figures in legal fees, maybe? If it goes for years?
A historic judgment against OpenAI (zero chance of that here) might actually drive them out of business, but they’re far beyond the threshold of being well-capitalized enough to fight Apple on this for years.
Huh what? Since when is OpenAI's IOU's and paper money worth more than Apple's cold hard cash? They have been backing away from their infrastructure commitments and literally pushing their IPO out.
Their "value" might fly with idiot VC's that are willing to throw money at them based on vibes. They are "big" on paper, nothing close to the financial heft of Apple. If anything this is where Sam Altman’s reality-distortion-field isn't going to fly.
> ... it results in these bizarro takes that OpenAI is finished as a company lol.
Meanwhile there's repeated news of OpenAI barely making ends meet and poised to run out of money mid next year. Unfortunately, OpenAI being 'well-capitalized' just doesn't hold up to what we've been actively seeing.
How much do you think they actually need to fight this?
BP spent a billion on defense attorneys for Deepwater Horizon. There’s a zero percent chance it costs OpenAI anywhere close to that to fight this, but even if it did, a billion over the next few years wouldn’t even be a speed bump for them.
This is specifically finance language for budgeting and one of the better ways to speak about this given the public disclosure of this information is substantial for their publicly traded stock.
This is exactly what the app does, except it ties it to agent activity and re-enables sleep once the turn ends, the laptop is close to overheating, the battery is low, or on a timer.
This works great. Only downside is display doesn't turn off when you close the lid. Made a little status bar utility to wrap the disablesleep command and address that downside:
I do exactly this (and more since my role is much broader and so is my approach) as a fractional head of product, data, and operations for multiple companies all in healthcare (fast growing self-funded to series D/IPOing soon). I saw your initial launch and felt validated by you all working on it, and now I’m further validated by the pivot. I have more work than I can handle, so I’m happy to share tips. You can find me via a bit of googling my HN handle or just adding a dot com to the end.
reply