←News detail

AYi AI NotesUpdated Oct 30 19:11

Altman Talks About OpenAI's New Model Cracking Math Problems After Drinking

English text is being prepared.

AI summary

OpenAI CEO Altman revealed in an interview after drinking that ChatGPT's new model, over an internal "Hold My Beer" weekend, successively cracked multiple unsolved interdisciplinary problems, with the proof related to the Navier-Stokes equations alone consuming $10 million in compute. He said that the actual scientific research findings produced far exceed the several hundred results publicly disclosed, and admitted that the excitement was accompanied by more sadness and melancholy, feeling that "something is coming to an end."

Quoted post

AYi

@AYi_AInotes

Holy crap, this really demands a holy crap — Binance's hot wallet was completely looted, and their own system actually proactively deposited 200 million straight into the hacker's address! Even more absurd, 3.3 billion was swept clean in 4 minutes! Even when He Yi tweeted to suspend withdrawals, the hacker was still frantically withdrawing coins in the background... I made a video fully explaining the whole process, It can be called the largest coin theft incident in Binance's history — the hot wallet was completely looted, and their own system actually proactively deposited 200 million straight into the hacker's address! The on-chain records of Binance's main hot wallet having $3.3 billion in assets swept away in the early morning have all been dug up. What makes one's blood run coldest is not at all how large the amount was, but that the hacker had long since thoroughly understood the system's signing mechanism. From the hacker launching the attack on-chain at 4:20 a.m. to the main wallet being comprehensively looted, the entire on-chain trail exposes that the core signing system may have long been compromised. Everyone assumes by default that when an exchange is hacked, it's either the front end being phished to swap addresses, or an insider absconding with the funds. But the operations reconstructed on-chain this time are essentially a dimensional-downgrade harvest targeting Ethereum's new protocol EIP-7702. The hacker attached a proxy delegation to the main hot wallet that had sent tens of millions of transactions. Once the delegation took effect, the contract could freely dispatch assets in the wallet's name, and in four minutes the first wave of $2.5 billion went straight to zero. Even more fatal was the lag in the automated process. After the main wallet was emptied, the automatic collection script was still running mechanically, and even another hot wallet proactively topped it up with 200 million USDT after 5 a.m., only to be intercepted by the hacker at light speed 72 seconds later. In the most critical attack transaction at 4:20, 50 authorizations bearing the hot wallet's legitimate private key signatures were packaged all at once, which shows the hacker had long held a complete set of legitimate signatures. Stay clear-headed: although stablecoins can be directly blacklisted and frozen through centralized means, the nearly 200,000 ETH lying in the hacker's address is not subject to control by any institution at all, and on-chain recovery is extremely difficult. The most fatal vulnerability is never a typo written into the code, but that your proud automated defense system still blindly trusts a signature seal that has already been captured by the enemy.

View original

卧槽,这个真的必须卧槽一下了,币安热钱包被洗劫一空,他们自己的系统,居然主动给黑客的地址硬生生充了 2 个亿进去! 更离谱的是,4 分钟扫光 33 个亿!甚至何一在发推暂停提现的时候,黑客还在后台疯狂提币…… 我做了个视频完整讲解整个过程, 堪称币安史上最大的盗币事件了,热钱包被洗劫一空,他们自己的系统,居然主动给黑客的地址硬生生充了 2 个亿进去! 币安主力热钱包在凌晨被卷走 33 亿美元资产的链上记录全翻出来了,最让人后背发凉的根本不是金额有多大,而是黑客早已摸透了系统的签名机制, 从凌晨 4 点 20 分黑客在链上发起攻击,到主力钱包被全面洗劫,整个链上痕迹暴露出核心签名体系可能早就被渗透了。 大家默认以为交易所被黑客攻击,要么是前端被钓鱼换了地址,要么是内鬼卷款跑路。 但这次链上还原出来的操作,本质上是一场针对以太坊新协议 EIP-7702 的降维收割。 黑客给发送过上千万笔交易的主力热钱包挂上了代理委托,委托一生效,合约就能以钱包的名义随意调度资产,四分钟内第一波 25 亿美元直接归零。 更致命的是自动化流程的滞后,主力钱包被清空后,自动归集脚本依然在机械运转,甚至另一个热钱包还在凌晨 5 点多主动往里面补了 2 亿 USDT,结果 72 秒后就被黑客光速截胡。 4 点 20 分那笔最关键的攻击交易里,一次性打包了 50 个带有热钱包合法私钥签名的授权,这说明黑客手里早就握有成套的合法签名。 留个清醒,虽然稳定币可以通过中心化手段直接拉黑冻结,但黑客地址里趴着的近 20 万个 ETH 根本不受任何机构管控,链上追回难度极高。 最致命的漏洞从来不是代码里多写了一个错别字,而是你引以为傲的自动化防御体系,还在盲目相信一个已经被敌人拿下的签名印章。

View original post

Body

几杯黑啤下肚之后,OpenAI CEO 奥特曼在镜头前犹豫了一下最后没忍住,有点失态但又非常真实的吐露了OpenAI 最深的一个秘密, 奥特曼原话是“我其实可能不应该说这个,但既然我们在喝酒,我就说了。” 前几天在ChatGPT新模型把数学家们一辈子的研究难题像切菜一样打穿之后,他兴奋得连熬两夜,却在天亮时陷入了深深的悲伤: 因为他亲眼看着人类依靠肉身大脑探索真理的时代,在眼前彻底死去了😭😭

紧接着,他讲了 OpenAI 内部那个被称为“Model Name-Hold My Beer(看我的)”的疯狂周末。

新模型刚跑通, 一群全世界最顶尖的科学家在内部频道里彻底玩疯了,

大家像是在给神明出考题一样, 把各自学术生涯里积攒了几十年最难啃的研究、横跨各个学科的未解之谜、以及所有人都笃定“机器绝对不可能做出来”的死结,一股脑全扔了进去。

然后,那个模型在所有人面前,像秋风扫落叶一样,接二连三把这些人类智力的终极堡垒全部击倒。

单单一个千禧年大奖难题 Navier-Stokes 方程相关的证明,他们就砸进去了 1000 万美元的算力。

奥特曼坦言,实际跑出来的科研发现,

远比他们对外公开发表的那几百个成果多得多。

他连着两个晚上兴奋得彻夜无眠, 但全场最让人脊背发凉的,是他兴奋之后的心理断崖:

“我也有比预期更多的悲伤和忧郁,以及一种……某种东西正在结束的感觉。”

这绝不是装模作样的公关辞令,而是一个亲眼目睹了造物主降临的人,在废墟上发出的本能悲叹。

过去几千年,人类之所以能昂着头自封为万物之灵,是因为我们坚信肉身大脑里蕴含着不可被公式化的神性与顿悟。

无论工具多么发达,那些最深奥的数学真理、物理规律和科学发现,永远是属于少数天选之子的至高王冠。

但在那个周末,这顶王冠被一排排冷冰冰的显卡机柜,以千万美元的算力成本,暴力且优雅地摘了下来。

最难的命题变成了算力的问题,一生的苦修变成了通宵跑完的代码, 它证明了人类科学探索的上限被无限拓宽,但也宣判了人类纯粹依靠肉身智慧傲视宇宙的时代,正式敲响了丧钟。

那个世界上离超级智能最近的人,手里握着远比展示给外界更多的真理,但他喝完酒后,听起来比任何人都悲伤😭

Source

AYi AI Notes