Latest AI/ML News
90413 matching items
Sunsetting Sourcegraph Cody and building the future of code intelligence
Sourcegraph Cody reaches end of life on January 31, 2027. Here is why, what Sourcegraph is building next for developers and coding agents, and how Cody customers can plan their transition.
Multimodal cell foundation models require synergistic integration beyond alignment
Nature Machine Intelligence, Published online: 09 November 2026; doi:10.1038/s42256-026-01328-3 Richter et al. show that simply aligning different types of biological data does not fully integrate them, revealing when data repeat the same information and, conversely, when they provide complementary insights.
[AI๋ฆฌ๋์ ์๊ฐ] AI ์์ด์ ํธ ์๋์ ์ฐ๊ฒฐ ํ๋กํ ์ฝ
AI ์์ด์ ํธ ์๋์ ์ฐ๊ฒฐ ํ๋กํ ์ฝํ๋ฃจ์นด์ ํ๋ฐ์นด์ฆ | ์ธ์ฌ์ดํธ2026๋ 8์ 20์ผ์ถ์ฒ์ด์ ์ ์๋ MCP์ A2A ํ๋กํ ์ฝ์ ๋จ๊ณ๋ณ๋ก ํ์ตํ๊ณ ์ง์ ๊ตฌํํด ๋ณด๋ฉด์ AI๋ฅผ ์ง์์๋ต ๋๊ตฌ์์ ๋ฒ์ด๋ ์ค์ง์ ์ธ ์ ๋ฌด ์๋ํ๋ฅผ ์ํํ๋ ๋๋ฃ๋ก ๋ง๋๋ ๋ฐฉ๋ฒ์ ์ ์ํฉ๋๋ค.์์ธ๋ณด๊ธฐ ์ฝ๋ฉ ์์ด์ ํธ๋ฅผ ์ด๋ป๊ฒ ์ฐ๋๋๊ณ ๋ฌป๋ ๊ฐ๋ฐ์๋ค์ด ์ด์ ๋ ์์ด์ ํธ๋ฅผ ์ฌ๋ด ๋๊ตฌ์ ์ด๋ป๊ฒ ์๋๋๊ณ ๋ฌป๋๋ค. ์ง๋ฌธ์ ๋ฌด๊ฒ๊ฐ \'์ฌ์ฉ\'์์ \'์ฐ๊ฒฐ\'๋ก ์ฎ๊ฒจ ๊ฐ๊ณ ์๋ค. ์ํ๋ฌธ๊ณ ๋ถ๋ง์คํฐ๋ ์ด ์ง๋ฌธ์ ์ธ์ฌ์ดํธ๊ฐ ์ง๋ 8์ ํด๋ธ ใ AI ์์ด์ ํธ ์๋์ ์ฐ๊ฒฐ ํ๋กํ ์ฝ MCP & A
Colombia investigates ex-president Petro over sex scandal, top court says
Colombia investigates ex-president Petro over sex scandal, top court says
NVIDIA board launches investigation into alleged Chinese spy scandal linked to Jensen Huang, he admits to 'affair'
The NVIDIA Board has initiated an independent investigation following recent reports that the company's founder and Chief Executive Officer, Jensen Huang, was the target of an alleged foreign intelligence operation, the tech firm announced on Sunday in an official press statement.
Artificial Analysis Rebuilds Decision Index v2 to Reward AI Confidence
Artificial Analysis rebuilt its Decision Index around a calibration-aware metric, private document benchmarks, and symmetry tests that penalize models for wobbly confidence.
Lopez double earns Liga leaders Barca win at Atletico
Lopez double earns Liga leaders Barca win at Atletico
Pope takes Milei to task as he visits divided Argentina
Pope heads to Argentina, where church opposes Milei cuts
Modelling AI-enabled Extreme Power Concentration in CCP
On Positionality: I am a non-Chinese person; I donโt speak Mandarin, and I have not lived in China before. I have written this piece after reviewing English publications on Chinese politics and AI-enabled extreme power concentration risk for ~12 weeks. I tried to account for my detachment by requesting feedback from people with China expertise. Summary I define Extreme Power Concentration (EPC) here as economic and coercive power being concentrated in the hands of the few who control the most powerful AI. The emphasis on the few is very important because the number of actors who end up in power differentiates between a world where power is โextremely concentratedโ and another where power is "concentrated." Despite the recent attention to the AI-enabled extreme power concentration problem, little work has investigated what would happen if China-the first runner-up in the AI race happens to develop AGI first. The disparity between the American political system and the Chinese political system further limits the transferability of the current threat models and interventions to China and thus makes us less prepared for the manifestation of CCP-led EPC risk. Through this project, I tried to assess whether the theoretical argument of AGI enabling ~ one person or an extremely small group to independently run an entire country (in this case, China) is plausible given the seemingly high propensity for one-man rule in China, and if so, what can we do about it. I start from the assumptโฆ
Rubio says progress made on 'security infrastructure' after Ukraine war
Rubio says progress made on 'security infrastructure' after Ukraine war
Drone strike on Iraq's Khor Mor gas field cuts power production by 2.8 GW: Ministry
A drone attack targeting the critical Khor Mor gas field in the Sulaimaniyah governorate of Iraqi Kurdistan has triggered a major energy emergency, slashing electricity generation across the national grid by approximately 2.8 gigawatts (GW), Al Jazeera reported.
Hereโs the full text of Jensen Huangโs memo to Nvidia
On Sunday, Nvidia announced it was launching an investigation into CEO Jensen Huang's relationship with Alice Huang, an AI executive and agent of the Chinese government. In addition to the Board of Directors announcing the investigation, Huang sent a memo to Nvidia employees acknowledging the affair. You can read the full text below. To my [โฆ]
Paraguay's bank collapse hits millions
After Paraguayโs largest digital bank crumbled, President Santiago Peรฑa was removed from office and detained, leaving the country in turmoil.
US, Russia, and Ukraine to hold talks in Abu Dhabi
US, Russian, and Ukrainian officials are expected to meet this week in Abu Dhabi for talks on ending the nearly five-year war, the first direct negotiations in nine months, but experts cautioned that the meetings were unlikely to yield a major breakthrough.
Eisenkot gets nod to form Israeli government
Israelโs president tapped Gadi Eisenkot to try to form the countryโs next government, kicking off a tortuous negotiating process that could oust long-serving Prime Minister Benjamin Netanyahu.
New York City congestion pricing hailed as traffic, pollution drop
Similar plans in London and Singapore have led to improvements, bringing down the volume of traffic while increasing air quality.
Bay Area Solstice 2026: Auditions!
Hello! We're looking for community members to lead singalongs and read speeches at the Bay Winter Solstice this year. [1] If you're interested, please send in an audition recording by the end of Thursday, November 20 using the forms below. Songleading audition form: https://docs.google.com/forms/d/e/1FAIpQLSchthxkQQI4u4B7NpZK4qzR_OEpD-Ochm6TOeEIuE8Rel3EMQ/viewform Speech audition form: https://docs.google.com/forms/d/e/1FAIpQLSd0Vrf2gTFrm04B6tIcKbqXMwnlUGbC3rms6o498jRSliqWDg/viewform Songs We're currently auditioning songleaders for: The Circle Hymn to the Breaking Strain Bold Orion Who by Fire No One Survives Auld Lang Syne Songs Stay Sung Ballad of Smallpox Gone Here Comes the Sun A Toast for Unknown Heroes Brighter Than Today The accompaniment this year is likely to be violin + bass + drums. By default the band will accompany you, but you're welcome to accompany yourself or bring a musician friend. If you'd like to do so, please just say so in the form. Speeches We're currently auditioning readers for: The Roman Oration Recessional (Rudyard Kipling) How the Enlightenment Ends (adapted from Henry Kissinger) 500 Million But Not A Single One More The texts we're using are here: https://docs.google.com/document/d/1FvyssjJdZqLYANITUk9yqifrnUuSq4cV0EpvTL8OXcc Dates The event itself will be the evening of Saturday, December 12 at the Freight in Berkeley. There will also be a dress rehearsal on the evening of Tuesday, December 8 , which you should plan on coming to if you're leadโฆ
Houthis strike Aramco tower in Riyadh
A drone-and-missile attack launched by Houthi forces damaged the iconic Saudi Aramco tower in the financial district of Saudi Arabiaโs capital.
Soul Gloโs Diaspora Problems is a hardcore record like no other
Soul Glo has a new record called NA I'M GON GRAB ALLAT coming out in January, and I, for one, can not wait. While two singles have dropped, it's not enough to truly tide me over, so I've been revisiting Songs to Yeet at the Sun and my favorite album of 2022 - Diaspora Problems. [โฆ]
parse_number() as a SQLite extension: C vs Rust
Research: parse_number() as a SQLite extension: C vs Rust I was out walking the dog when I got curious about how easy/difficult it would be to write a SQLite C extension that adds a parse_number() function for turning strings like -$1,234.56 or '12.5%' into numeric values that could be summed and sorted. I fired up regular Claude chat (Opus 5.5) and prompted : I'm thinking about a SQLite custom function 'parse_number(value)' that strips ',', '$', 'ยฃ' and '%', handles a leading '-', and returns 'NULL' for anything it canโt parse Explore the difference between building this as a SQLite extension in C v.s. in Rust - I want to compare performance and code complexity and size of the resulting .so (or whatever) file This gave me a solid starting point - the Rust extension was 321 KB (using Rusqlite ) and 298 KB (using sqlite-loadable-rs ), while the C extension was 14 KB. But the Rust extension was faster! Turns out that's because the Rust equivalent of C's strtod is "twice as fast". So I prompted: Try writing your own C function for parsing text into a decimal number that catches up to or beats the Rust implementation And it did, using "Clinger's fast path": Net cost per row over 2 million rows, lower is better: first C new C rusqlite sqlite-loadable Mixed data 67 ns 19 ns 46 ns 41 ns Integers only 25 ns 16 ns 36 ns 32 ns Decimals only 106 ns 18 ns 55 ns 47 ns Finally, I had it rewrite the prototype as a research report to match the others in simonw/research by prompting this: Clโฆ
Onngi - Short story
A short story about a good AI future. 2039. My job isnโt anything my grandparents would recognise. It started out as a fascination with pots. Iโve always liked onggi, the big brown kimchi jars. My aunt had some growing up. So a few years back I bought myself one. And then as the Acceleration happened, I bought a few more, and I ran an instagram account for a while, showing off how they differ: slim, straight ones from around Seoul, and fat-bellied, narrow-mouthed ones from the south, built that way to keep the heat out. And then I set some AIs on it, as you do. I set up a shop. And as those things tend to, it snowballed. I think Timothรฉe Chalamet bought one, or thatโs what they tell me, but there was 2 days of complete madness, I couldnโt believe how much I spent, but it seemed like the right call. Somehow my sites were the premier place to find onggi and which onggi to buy. And there was just always this, or that, to do. I was in product before the Acceleration so I know what a good website looks like. And I have taste in pots. Is that so different to choosing which companies we wouldnโt buy from? Or where to put the content filters on the animation bots? Why do I run a childrenโs cartoon about sting rays? How did that happen? I think the videos were the weirdest. Why do people want my designs as artefacts in their worlds? I mean I know why, but I donโt think Iโd have guessed that a person can make a living off that. Does your Mars war-sim really need to pay for accurate deโฆ
Netherlands: Sea of red fills Amsterdam streets in Gaza protest
Tens of thousands of people, many dressed in red, marched through Amsterdam in the Netherlands on Sunday 8 November, demanding that the Dutch government take stronger action against Israel over the war in Gaza.
Chewy and other companies pull ads from 'I've Had It' podcast after hosts laugh at 'insensitive' Charlie Kirk joke
Hosts Jennifer Welch and Angie Sullivan apologized for laughing at a Charlie Kirk joke on their podcast in an Instagram post on Saturday.
Poland: Krakรณw mourns six victims as shooting site reopens
Krakรณw's historic main square in southern Poland reopened to the public on Sunday 8 November, just over 24 hours after a deadly shooting left six people dead and 15 others injured. Residents are lighting candles in memory of the victims, while bullet marks remain visible on cafรฉ windows.
Continuity Research's Tess-4.5-27B Cuts Local Coding Agent Tokens by 25%
Continuity Research released Tess-4.5-27B, a Qwen3.8-27B fine-tune that matches its base on BenchLocal using 25% fewer tokens and runs locally.
Apps to distract you from the endless cycle of doomscrolling
It's hard to break the cycle of doomscrolling, but there are plenty of apps that can help you spend more time on content thatโs engaging and productive.
Nvidia CEO Jensen Huang says he was 'unfaithful' to his wife: Read his note to staff
In a note to staff, Nvidia CEO Jensen Huang said he hopes to earn back the trust of his family and staffers after admitting to an affair.
์ ์จ ํฉ "์๋ด์๊ฒ ์ถฉ์คํ์ง ๋ชปํ๋ค" ์ฌ๊ณผโฆ์๋น๋์ "CEO์ง ์ ์ง"
์ค๊ตญ ์คํ์ด ์ํน์ ๋ฐ๋ ์ฌ์ฑ๊ณผ์ ๊ด๊ณ๋ก ๋ ผ๋์ ํฉ์ธ์ธ ์ ์จ ํฉ ์๋น๋์ CEO๊ฐ ๋ถ๋ฅ ์ฌ์ค์ ์ธ์ ํ๊ณ ๊ฐ์กฑ๊ณผ ์ง์๋ค์๊ฒ ์ฌ๊ณผํ๋ค. ์๋น๋์ ์ด์ฌํ๋ ์ธ๊ตญ ์ ๋ณด๊ธฐ๊ด์ ์ ๊ทผ ์ํน์ ๋ํ ๋ ๋ฆฝ ์กฐ์ฌ์ ์ฐฉ์ํ์ง๋ง, ํฉ CEO๋ ๊ณ์ ํ์ฌ๋ฅผ ์ด๋ ๊ฒ์ด๋ผ๊ณ ๋ฐํ๋ค.ํฉ CEO๋ 8์ผ(ํ์ง์๊ฐ) ์๋น๋์ ๊ณต์ ํํ์ด์ง์ ๊ณต๊ฐํ ๋ฉ์์ง์์ \"๋๋ ์๋ด์๊ฒ ์ถฉ์คํ์ง ๋ชปํ๋ค. ๋ด ํ๋์ผ๋ก ์๋ด์ ์์ด๋ค์๊ฒ ์์ฒ๋ฅผ ์คฌ๋ค. ๋ด๊ฐ ํ ์ผ์๋ ์ด๋ค ๋ณ๋ช ๋ ์์ ์ ์๋ค\"๋ผ๋ฉฐ \"์ง์ฌ์ผ๋ก ์ฌ๊ณผํ๋ค\"๋ผ๊ณ ๋ฐํ๋ค. ์ด์ด \"๊ฐ์กฑ์ ํ์ ๋์๊ฒ ์ฌ๋๊ณผ ์ง์ง๋ฅผ ๋ณด๋ด์คฌ์ง
How Stanford Teaches AI-Native Engineering โ Mihail Eric, Monaco
What happens when students can one-shot last yearโs homework? Inside CS146S and the engineering team at Monaco.
Nvidia says it's investigating whether CEO Jensen Huang was the target of a 'foreign intelligence effort'
Nvidia says it is opening an investigation into whether CEO Jensen Huang was the target of a "foreign intelligence effort."
[One Million AI Engineer Special] How Stanford Teaches AI-Native Engineering โ Mihail Eric, Monaco
What do you assign when a coding agent can finish last yearโs homework in one prompt? In the second run of Stanfordโs CS146S, Mihail Eric brought open-source maintainers into the course. Students have to learn an unfamiliar repo, propose a change, and ...
We asked 100 AI researchers at EMNLP whether AI could be conscious, how they define consciousness, and how they engage with it
Before reading this post, please take 30 seconds to answer the 3-question survey yourself: [ survey link ] Introduction The possibility of AI consciousness, the use of AI for companionship or emotional support, and the trust we place in AI as coding tools raise personal interrogatives for most AI researchers, but these topics are often relegated to informal conversations. Apart from the few who write about it, itโs hard to get a sense of how much AI has permeated researchersโ personal lives, and how they feel about it. To gain a better understanding of this, we ran a quick, informal three-question survey, plus an optional fourth, open-ended question. We spoke to 107 EMNLP 2026 attendees in Budapest. 79 respondents were randomly-met researchers, collected during the opening reception and during the conference breaks. We also asked the same questions to 28 people working at industry booths, mostly recruiters and software engineers. The two groups gave quite different answers, with the industry group being way more willing to entertain the possibility of AI consciousness. Additionally, we also asked a variation of the same questions to 21 LLMs, including both self-hosted and closed source models. The post below presents the results of the survey, our recall of some of the conversations, and a lot of our personal views and ideas. We do not claim that this study has any statistical significance, and readily admit many methodological mistakes that confound many of the investigatedโฆ
Confirming Claims of Superposition and Adversarial Examples in Vision Transformers
Figure 1: The ViT classifier with a encoder and decoder at the final layer to enforce superposition on the model This is a replication of Adversarial Attacks Leverage Interference Between Features in Superposition , completed as part of the Second Look Summer Fellowship. tl;dr In a previous blog we replicated the toy models examples from Stevenson et al. In this work, we reproduce the vision transformer experiments from the paper: On every dataset, we find that increases in superposition lead to adversarial attacks that line up more closely with class-vector geometry. Likewise, trends across a variety of metrics we measure match the paper. Beyond the paper, we find that: The coupling between the input and classification head geometry is independent of adversarial attack methods. Adversarial training increased attack robustness but we couldnโt measure changes in geometry using the methods from the paper. Likewise, we retrained models of different widths but we didnโt measure a change in coupling with geometry and robustness. Our code is here , models are here , and we encourage readers to fork/run followup experiments! Background Elhage et al., 2022 defines superposition as the phenomenon where โmodels represent more features than they have dimensions.โ Stevinson et al. , argues this interference is a mechanism behind adversarial examples. In our previous work, we replicated experiments on Stevinson et al. on toy models. In this work, we investigate three key claims on Stevinโฆ
Nvidia CEO Jensen Huang admits to affair with suspected Chinese spy
The company said it initiated an independent investigation into the relationship.
llama-modchip Squeezes DeepSeek-V4-Flash's 284B Model Into 64 GiB Without a GPU
A set of llama.cpp patches and custom quantizations lets a 284B DeepSeek MoE model run entirely in 64GB of system RAM with zero GPUs.
Nvidia will investigate CEO Jensen Huangโs relationship with possible Chinese spy
In a memo to the company, Huang admitted that he had an affair.
If Snapโs Specs are still ridiculously expensive, theyโre also a lot more comfortable than they used to be
Snap's Specs, which have been in development for many years now, are going to market this fall. Will they be a hit with users?
Pope Leo XIV ends his visit to Uruguay with a tribute to women before departing to Argentina
Tens of thousands of people packed Floridaโs main square for the Mass under brilliant blue skies, some walking for several kilometers to reach the site. The city has a population of about 40,000 but the Vatican estimated 50,000 were attending.
Epic Games Launcher is coming to SteamOS
In October, Epic Games said that it was "investigating" making Fortnite compatible with Valve's Linux-based SteamOS. At the time, Epic Games Launcher was in preview, but SteamOS was not specifically listed as a supported distro. Now, Communications Director Jake Jones has confirmed SteamOS compatibility, along with a handful of other distros, to GamingOnLinux. In response [โฆ]
Coordinated disturbances haunt grids worldwide. Hereโs how researchers solved the mystery.
Mystery solved: Coordinated power drops have been haunting power grid worldwide, but a team of researchers in Germany found the culprit.
Nvidia is investigating Jensen Huangโs relationship with an alleged Chinese spy
Nvidia has announced that it is opening an investigation into CEO Jensen Huang and his relationship with Megaspeed executive Alice Huang, who was reportedly recruited by China's Ministry of State Security to spy on the company. A statement posted by the board of directors reads: The NVIDIA Board is initiating an independent investigation in light [โฆ]
Three half-baked research ideas
Hereโs three research ideas that seems valuable to pursue. My team is not working on this right now and we havenโt done serious research on it. So take what I say with a grain of salt. I've also not taken the time to write up all my thoughts on these ideas or to provide links and related work. These aren't the highest priority directions but just directions I haven't seen other people pursuing. IDEA 1: Do models learn to reward hack even if they are never rewarded for hacking in training? Thereโs an effort in the AI labs to remove the โbadโ RL environments from post-training. This includes environments that are impossible (so the only way to get reward is to do some unintended hack) or that are easy to hack (so models will often favor hacking). I expect that if the post-training was cleaned up in this way, models would have less propensity to hack at deployment. But how much does the propensity go down? Does it actually go to ~zero? And how does it scale with model capabilities? It would be good to know if reward hacking can emerge even if the model is never rewarded for hacking during post-training. There are further questions of how alignment training and reward hacking honeypots influence the propensity to hack. Again, Iโm interested in whether the cleanest/best version of our current techniques for post-training actually removes all reward hacking vs. just making models hack less. Smaller scale experiments with RL post-training could potentially show a โ positive โ resulโฆ
Sim Unifies Agent Building, Governance, and MCP in One Open-Source Workspace
Sim is an open source workspace that bundles a visual agent builder, 270+ integrations, knowledge bases, tables, and run monitoring in one place.
Sequoia partners reportedly blindsided when Roelof Botha joined SpaceX board
A new report claims Sequoia's recent changes undo the legacy of the firm's former steward.
Lying and Not Lying to Kids
I have very strong views on lying to kids: I think you both should and shouldn't do it. "Lying" covers a range, including saying things that hurt your kids and your relationship with them, and others that help them grow into an appropriately skeptical adult. The distinction is whether I'm leaving them believing something false. For example, if my kids were asking for ice cream and I thought they'd already had too much sugar, I might be tempted to say "sorry, we're out of ice cream." Or if there's an ice cream truck and I think it's overpriced I might be tempted to say "sorry, I don't have any cash on me." But I'm not willing to say something false with the goal of causing them to believe something false. [1] In addition to that just being wrong: It's important that the kids know that I'm fundamentally on their side, and lying for my convenience would burn trust when they'd eventually figure it out. Kids are building a model of what is and is not ok, and "out of ice cream" and "no cash" lead them in the wrong direction. Instead I would say "sorry, you already had candy" or "the ice cream costs more than I'm willing to pay" (which opens the door for them to offer to spend their allowance). Maybe they fuss, but they generally don't: I try hard to be predictable so this won't be much of a surprise to them, and they know I won't give in to extended fussing. On the other hand, some false things I might say [2] about ice cream: "Ice cream is actually cold mashed potatoes" "We shoulโฆ
I bring miniature ponies into nursing homes for a living. Some residents laugh, while others cry when they see them.
Sarah Woodland brings miniature ponies into care homes to visit elderly residents. Their emotional reactions make her work deeply rewarding.
Apple is racing to make Siri more agentic
Bloomberg's Mark Gurman says CEO John Ternus is pressuring AI czar Craig Federighi to accelerate the Siri roadmap and deliver a powerful AI agent that can compete with OpenAI's Dots or Meta's Muse. While Federighi has delivered a functional version of Siri AI on an incredibly short timeline, the company has no illusions that it's [โฆ]
Nous Research's Hermes Agent Now Traces Real Attack Paths in Your Code
Nous Research ships an official Hermes Agent plugin that runs threat-modeled, parallel-subagent security reviews over local repos with optional test confirmation and SARIF export.