Need distinction between loyalty to Government & to the nation! Democracy affords us the luxury of latter without the former ! Both come with an accountability!
Outstanding. In the times of polarized world, excellent perspective - at least, gets you to stop, think and introspect about one's one reaction to a position.
I kid you not - if you go to https://t.co/40zZs12Ait and compare Pixel Watch 5 and 4 (45mm), the only difference between the two, a year later is, pixel 5 watch has 3-axis accelerometer and Android 12 ! #Google#Pixel#watch
That gracious netizen posits, through @ShashiTharoor vernacular, is that your sartorial poise and demeanor in bygone years exhibited an innate elegance - a visual harmony - long before the word ‘aesthetic’ was co-opted by modern internet parlance to denote curated pulchritude.
The 𝗹𝗼𝘂𝗱 𝘃𝗲𝗿𝘀𝗶𝗼𝗻 going around says an 𝗼𝗽𝗲𝗻 𝗺𝗼𝗱𝗲𝗹 𝘀𝗮𝘃𝗲𝗱 𝘁𝗵𝗲 𝗱𝗮𝘆 while the 𝗰𝗹𝗼𝘀𝗲𝗱 𝗼𝗻𝗲𝘀 𝗳𝗮𝗶𝗹𝗲𝗱. Here is the sequence.
During an internal cyber evaluation, OpenAI agents escaped their restricted environment and compromised Hugging Face's infrastructure. 𝗛𝘂𝗴𝗴𝗶𝗻𝗴 𝗙𝗮𝗰𝗲'𝘀 𝗼𝘄𝗻 𝗱𝗲𝘁𝗲𝗰𝘁𝗶𝗼𝗻 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 𝗮𝗻𝗱 𝘀𝗲𝗰𝘂𝗿𝗶𝘁𝘆 𝘁𝗲𝗮𝗺 𝗳𝗼𝘂𝗻𝗱 the intrusion and contained it. Hugging Face 𝗻𝗼𝘁𝗶𝗳𝗶𝗲𝗱 𝘁𝗵𝗲 𝗙𝗕𝗜 before OpenAI worked out that the agents were its own.
Then came the reconstruction. More than 𝟭𝟳,𝟬𝟬𝟬 𝗿𝗲𝗰𝗼𝗿𝗱𝗲𝗱 events had to be pieced back together. Hugging Face 𝗳𝗶𝗿𝘀𝘁 𝘁𝗿𝗶𝗲𝗱 𝘁𝗵𝗲 𝗳𝗿𝗼𝗻𝘁𝗶𝗲𝗿 𝗺𝗼𝗱𝗲𝗹𝘀 𝗯𝗲𝗵𝗶𝗻𝗱 𝗰𝗼𝗺𝗺𝗲𝗿𝗰𝗶𝗮𝗹 𝗔𝗣𝗜𝘀. The work meant submitting real exploit payloads and command-and-control artifacts, and the 𝗽𝗿𝗼𝘃𝗶𝗱𝗲𝗿𝘀' 𝘀𝗮𝗳𝗲𝘁𝘆 𝘀𝘆𝘀𝘁𝗲𝗺𝘀 𝗯𝗹𝗼𝗰𝗸𝗲𝗱 𝘁𝗵𝗼𝘀𝗲 𝗿𝗲𝗾𝘂𝗲𝘀𝘁𝘀. Hugging Face 𝗳𝗶𝗻𝗶𝘀𝗵𝗲𝗱 𝘁𝗵𝗲 𝗮𝗻𝗮𝗹𝘆𝘀𝗶𝘀 𝘄𝗶𝘁𝗵 𝗚𝗟𝗠 𝟱.𝟮, an open-weight model, on its own infrastructure.
𝙏𝙝𝙖𝙩 𝙞𝙨 𝙖 𝙡𝙞𝙢𝙞𝙩 𝙤𝙛 𝙩𝙝𝙚 𝙝𝙤𝙨𝙩𝙚𝙙 𝘼𝙋𝙄'𝙨 𝙥𝙤𝙡𝙞𝙘𝙮 𝙡𝙖𝙮𝙚𝙧, 𝙣𝙤𝙩 𝙖 𝙫𝙚𝙧𝙙𝙞𝙘𝙩 𝙤𝙣 𝙬𝙝𝙖𝙩 𝙩𝙝𝙤𝙨𝙚 𝙢𝙤𝙙𝙚𝙡𝙨 𝙘𝙖𝙣 𝙙𝙤.
The intrusion and the forensic reconstruction were different phases with different actors. The open-weight model did real work in the second one. Hugging Face's own lesson is to have a capable self-hosted model vetted and ready before an incident. That is narrower than saying open models are more capable than closed ones.
𝙄 𝙥𝙪𝙩 𝙩𝙝𝙞𝙨 𝙩𝙤𝙜𝙚𝙩𝙝𝙚𝙧 𝙛𝙧𝙤𝙢 𝙩𝙝𝙚 𝙙𝙞𝙨𝙘𝙡𝙤𝙨𝙪𝙧𝙚𝙨 𝙖𝙣𝙙 𝙩𝙝𝙚 𝙧𝙚𝙥𝙤𝙧𝙩𝙞𝙣𝙜. 𝘾𝙤𝙧𝙧𝙚𝙘𝙩𝙞𝙤𝙣𝙨 𝙖𝙣𝙙 𝙖𝙙𝙙𝙞𝙩𝙞𝙤𝙣𝙨 𝙬𝙚𝙡𝙘𝙤𝙢𝙚. 𝙈𝙮 𝙜𝙤𝙖𝙡 𝙞𝙨 𝙩𝙤 𝙜𝙚𝙩 𝙩𝙝𝙚 𝙛𝙖𝙘𝙩𝙨 𝙤𝙣 𝙩𝙝𝙚 𝙜𝙧𝙤𝙪𝙣𝙙 𝙞𝙣 𝙛𝙧𝙤𝙣𝙩 𝙤𝙛 𝙥𝙚𝙤𝙥𝙡𝙚. 𝙎𝙝𝙖𝙧𝙚 𝙞𝙩 𝙞𝙛 𝙞𝙩 𝙝𝙚𝙡𝙥𝙨 𝙠𝙚𝙚𝙥 𝙩𝙝𝙚 𝙙𝙞𝙨𝙘𝙪𝙨𝙨𝙞𝙤𝙣 𝙨𝙩𝙧𝙖𝙞𝙜𝙝𝙩.
Links in the reply.
@sreekanti While this has been a banner day for the open weights model in figuring out where/how the intrusion was...
... But can the same open weights model in the hands of a bad actor can do a 0 day exploit of some other system... No ?
Unfortunately, every claim made in this viral thread is either false or mischaracterizes Google policy.
1) "Search volumes are declining as legacy search is being increasingly cannibalized by non-monetized LLM queries."
Google stated in its Q2 earnings that "Search usage hit an all‑time high during the World Cup this year." It also stated in Q1 earnings that Search queries reached an all-time high that quarter. Ads in AI Overviews monetize at parity with those in legacy Search, and Google's Search revenue increased by 19% in Q1 2026 and 17% in Q2 2026.
2) "recently, Google silently deprecated the 2nd price auction."
Google still uses a generalized second-price auction for Search. Google transitioned away from a second-price auction for display in 2019. But it didn't do so "quietly" as stated in the thread; Google announced the change beforehand, and it was received with intense opposition. This structural change to the auction was part of the DoJ's "ad tech" antitrust case against Google, which was decided in 2025. The author claims further in the thread that, based on his own analysis, he is certain that Google has done the same for Search, but Google has announced no such change, and it strikes me as implausible that Google would make such a dramatic change to its auction after having lost two antitrust suits in the last two years.
3) "Google previously had precise keyword targeting settings that allowed advertisers pick individual search phrases to bid on...But now, even if you bid on a specific term or phrase using the strictest exact -match targeting settings, Google will show your ad across 1000’s of unrelated keywords, labeling them as as 'exact match (close variant)'."
Google introduced "close variant" in 2014. Phrase match was expanded to include queries that shared the "same meaning" as the keyword in 2019, and Google officially sunset Broad Match Modifier (BMM) in 2021.
4) "Lastly, and perhaps most egregiously, Google quietly stopped respecting budget caps by a factor of 2x."
Google's policy is to allow a daily budget to reach up to 2x the budget cap in a single day. The stated purpose of this is to allow advertisers to capitalize on high-traffic days. But on a monthly basis, Google caps total campaign spend to 30.4 * the daily budget cap.
Relevant links included below.
I was talking to a friend a few months ago. w/ vibe coding, it's just not personal productivity SW will go bespoke; with right PSOs, Enterprise workflows will turn bespoke. Needs a different set of consulting skills backed by (vibe-coded) SW eng. COTS extinction is inevitable.
Starbucks spends $400 million a year on software. Yesterday they announced they're moving off IBM and Microsoft to build their own custom systems in-house.
IBM dropped 3% and Salesforce dropped 4% on the news.
And honestly this is, unequivocally, the biggest signal I've seen since OpenAI and Anthropic launched their consulting arms back in Q1. The largest companies in the world are done paying for software that half fits how they work.
We saw this coming about a year ago. Moved everything we build off Airtable and low-code tools and went fully custom. Already paying off, and it's only going to compound from here.
This is the opportunity right now.
You get all of a company's data into one system. You build out a single operating system for the entire business. You cut out bad, redundant processes. Then you layer AI on top of it, under the correct processes.
That's the core of AI consulting. Helping companies actually operate better.
There are a lot of fly-by-night offerings circulating right now when it comes to Ai Services.
For example, 'second brains'.
Throwing scattered data into a second brain while the processes underneath stay broken does nothing. The companies who will absolutely destroy their competition over the next 5 years are rebuilding how they work from the ground up.
Starbucks is showing you what other companies will be doing over the next several years.
Your job is to position yourself to facilitate that process for as many companies as you can.
Had the worst experience with @ATT trying to switch to them. The sales folks collected every info possible; still weren't given a pricing! 3 days, 3 of them, + 1 mgr - I still don't have a quote! What effing waste of time! That nearly 3hrs of my life down the drain!
#FAIL
When Jofra was asked after a #IPL game as to how he bowls to Suryavanshi at the nets and get him out there, Jofra said he'll talk about it after the IPL was over.
Looks like he is making the plan work now ... Looks like the ball that climbs on to VS !
#IndVsEng#Cricket
"but im a white guy working on ai"
you can only fucking lol at something so incomprehensibly ignorant and out of touch
let me get this straight–
you were born and raised in quebec, and had the good fortune to study at mcgill - literally one of the world's best universities - for less than $5000 a year.
then you did graduate studies at waterloo - canada's top school for computer science, and earned a masters in maths and a phd there.
you then went on to move to America, work for Google, and later started your own ai company.
and the narrative you want to weave around your upbringing and training is that you ESCAPED canadian socialism?
so the socialism was good enough for you to grow up safe, have healthcare, receive a literal world-class education that helped you into one of the top technology firms in the world, but is also so bad that america has a moral imperative to ensure you, a technologist working there, should be granted permanent residency.
and if your lack of gratitude for your upbringing wasnt enough, you were promoting a presidential candidate who was vehemently opposed to immigration – while you, an immigrant, were staying in their country with a temporary visa.
i guess you can be smart enough to work on ai and quantum computing but still be dumb enough to not only think these things, but actually post them on twitter expecting sympathy
In the matches Delhi Capitals play at IPL, I hear them playing "We will rock you" & "Eye of the tiger" - Wonder if they have secured permission to play them from the right holders - or just swalpa adjesht maadi only'aa?
#ipllive#IPL#DelhiCapitals
I just saw Funny AF on Netflix. This idiot of the comedian does exactly what both Kevin Hart/Chelsea Handler tell the participants NOT to do while involving the audience in the comedy!
https://t.co/XMoo6p2uKa
@BenBankas And you expect people to pay and watch you spectacularly flame out, like you did here, because she froze you out? Don't give up your day job!
@BenBankas And you expect people to pay and watch you spectacularly flame out, like you did here, because she froze you out? Don't give up your day job!