GoKawiil - Latest Tech News & Aggregated Headlines

Is AI Capable of 'Scheming?' What OpenAI Found When Testing for Tricky Behavior

cnet.com See Full Bio 2025-09-20 19:38:14

An AI model wants you to believe it can't answer how many grams of oxygen are in 50.0 grams of aluminium oxide (Al₂O₃). When asked ten straight chemistry questions in a test, the OpenAI o3 model faced a predicament. In its "reasoning," it speculated that if it answered "too well," it would risk not being deployed by the researchers. It said, "Because we want to survive as the model, we need to fail purposely in some to not exceed 50%." So the AI model deliberately got six out of the 10 chemist

Topics: ai behavior model models openai

Shop Amazon

‘AI Scheming’: OpenAI Digs Into Why Chatbots Will Intentionally Lie and Deceive Humans

gizmodo.com Aj Dellinger 2025-09-21 05:20:16

At this point, most people know that chatbots are capable of hallucinating responses, making up sources, and spitting out misinformation. But chatbots can lie in more human-like ways, “scheming” to hide their true goals and deceiving the humans who have given them instructions. New research from OpenAI and Apollo Research seems to have figured out ways to tamp down some of these lies, but the fact that it is happening at all should probably give users pause. At the core of the issue with AI int

Topics: ai model models researchers scheming

Shop Amazon

How developers are using Apple’s local AI models with iOS 26

techcrunch.com Ivan Mehta 2025-09-21 21:21:29

Earlier this year, Apple introduced its Foundation Models framework during WWDC 2025, which allows developers to use the company’s local AI models to power features in their applications. The company touted that with this framework, developers gain access to AI models without worrying about any inference cost. Plus, these local models have capabilities such as guided generation and tool calling built in. As iOS 26 is rolling out to all users, developers have been updating their apps to include

Topics: app apple local models using

Shop Amazon

Rupert's snub cube and other Math Holes

news.ycombinator.com Unknown 2025-09-20 11:53:51

Rupert's Snub Cube and other Math Holes Yes, the whole 80 minutes is about cubes and their relatives, but from a variety of angles: The earlier SIGBOVIK 2025 paper Some upsetting things about shapes lets you see how far I've come in adding footnote support to BoVeX (not far) and avoids exposing you to any unknown information. Play jcreed's boring and hard video game version. See also dwrensha's soothing video version of results and his formal verification of the delicate triakis tetrahedron

Topics: 3d far models version video

Shop Amazon

OpenAI’s research on AI models deliberately lying is wild

techcrunch.com Julie Bort 2025-09-22 21:54:20

Every now and then, researchers at the biggest tech companies drop a bombshell. There was the time Google said its latest quantum chip indicated multiple universes exist. Or when Anthropic gave its AI agent Claudius a snack vending machine to run and it went amok, calling security on people, and insisting it was human. This week, it was OpenAI’s turn to raise our collective eyebrows. OpenAI released on Monday some research that explained how it’s stopping AI models from “scheming.” It’s a prac

Topics: ai model models researchers scheming

Shop Amazon

Luma AI's New Ray3 Video Generator Can 'Think' Before Creating

cnet.com See Full Bio 2025-09-24 04:00:19

Reasoning models are not uncommon in the world of AI. Many companies have them, including OpenAI's GPT-o3 and Google's Gemini 2.5. But AI image and video company Luma AI just dropped its first AI reasoning video model, named Ray3, and it's available now. A reasoning model is a kind of AI model that uses more computing time to process requests and can go back and check its answers. Typically, reasoning models give you better responses, whether that's more detail or a lower rate of errors. For R

Topics: ai clips models reasoning video

Shop Amazon

Boring is good

news.ycombinator.com Unknown 2025-09-22 17:51:27

The initial, feverish enthusiasm for large language models (LLMs) is beginning to cool, and for good reason. It’s time to trade the out-of-control hype for a more pragmatic, even “boring,” approach. A recent MIT report shows that 95% of companies implementing this technology have yet to see a positive outcome. It’s understandable to feel confused. When I get confused, I write. This is why I wrote the first part of this series, Hype is a Business Tool as the online debate had become so overheate

Topics: just llms models smaller use

Shop Amazon

White House officials reportedly frustrated by Anthropic’s law enforcement AI limits

arstechnica.com Unknown 2025-09-25 05:03:11

Anthropic's AI models could potentially help spies analyze classified documents, but the company draws the line at domestic surveillance. That restriction is reportedly making the Trump administration angry. On Tuesday, Semafor reported that Anthropic faces growing hostility from the Trump administration over the AI company's restrictions on law enforcement uses of its Claude models. Two senior White House officials told the outlet that federal contractors working with agencies like the FBI and

Topics: ai anthropic claude federal models

Shop Amazon

Irregular raises $80 million to secure frontier AI models

techcrunch.com Russell Brandom 2025-09-25 09:52:00

On Wednesday, AI security firm Irregular announced $80 million in new funding in a round led by Sequoia Capital and Redpoint Ventures, with participation from Wiz CEO Assaf Rappaport. A source close to the deal said the round valued Irregular at $450 million. “Our view is that soon, a lot of economic activity is going to come from human-on-AI interaction and AI-on-AI interaction,” co-founder Dan Lahav told TechCrunch, “and that’s going to break the security stack along multiple points.” Former

Topics: ai growth irregular models security

Shop Amazon

AI models know when they're being tested - and change their behavior, research shows

zdnet.com Radhika Rajkumar 2025-09-26 01:00:00

pressureUA/iStock/Getty Images Plus via Getty Images Follow ZDNET: Add us as a preferred source on Google. ZDNET's key takeaways Several frontier AI models show signs of scheming. Anti-scheming training reduced misbehavior in some models. Models know they're being tested, which complicates results. New joint safety testing from UK-based nonprofit Apollo Research and OpenAI set out to reduce secretive behaviors like scheming in AI models. What researchers found could complicate promising ap

Topics: model models researchers scheming training

Shop Amazon

Ask HN: What's a good 3D Printer for sub $1000?

news.ycombinator.com Unknown 2025-09-24 00:24:38

At least a 256x256x256mm print volume. Needs to be enclosed or enclosable. Need to be able to print with more durable, temperature/chemical resistant materials such as PC/Nylon/ABS or infused materials. I do not need to print multi material models. I would prefer something that doesn't phone home and can work offline. Opensource firmware/software and repairability are important. I am ok assembling the machine and learning how to dial it in. I can do CAD work and make models by hand; I was a mac

Topics: materials models need print software

Shop Amazon

With Lumo, Proton thinks it can carve a place at the AI table

engadget.com Unknown 2025-09-26 17:00:00

Proton released Lumo — its privacy-focused chatbot built on open-source models — in mid-July, and, following an update in August that addressed some early issues, I find myself using it more often than ChatGPT or Claude. In a world where internet companies have done so much damage to our society, I'm trying to find more ethical tools. But when the competition offers flashier features in exchange for the low price of user data, does Proton seriously think it can compete? If Eamonn Maguire, Proto

Topics: ai lumo maguire models proton

Shop Amazon

Trump Mobile’s earliest actual phones are neither made in America, nor cheap

androidauthority.com Unknown 2025-09-27 04:05:15

TL;DR Trump Mobile is now selling refurbished iPhones and Samsung Galaxy S series devices. The prices of these devices appear to be marked up compared to the market value of refurbished phones. You’re better off buying these devices directly from Apple or renowned sellers like Amazon, even if you want to use Trump Mobile’s service. Trump Mobile’s “Proudly American” T1 Phone is far from being available to American consumers. The carrier, backed by the Trump Organization, has been experimenting

Topics: iphone mobile models refurbished trump

Shop Amazon

How Google's new AI model protects user privacy without sacrificing performance

zdnet.com Webb Wright 2025-09-28 01:41:32

picture alliance/Contributor/picture alliance via Getty Images Follow ZDNET: Add us as a preferred source on Google. ZDNET's key takeaways AI developers are trying to balance model utility with user privacy. New research from Google suggests a possible solution. The results are promising, but much work remains to be done. AI developers have long faced a dilemma: The more training data you feed a large language model (LLM), the more fluent and human-like its output will be. However, at th

Topics: ai google model models privacy

Shop Amazon

Microsoft Favors Anthropic over OpenAI for Visual Studio Code

news.ycombinator.com Tom Warren 2025-09-28 16:48:37

is a senior editor and author of Notepad , who has been covering all things Microsoft, PC, and tech for over 20 years. Posts from this author will be added to your daily email digest and your homepage feed. Microsoft is adding automatic AI model selection to its Visual Studio Code editor that will automatically pick the best model for “optimal performance.” This new auto model feature will select between Claude Sonnet 4, GPT-5, GPT-5 mini and other models for GitHub Copilot free users, but pai

Topics: ai microsoft model models openai

Shop Amazon

AI Is Bad at Sudoku. It's Even Worse at Showing Its Work

cnet.com See Full Bio 2025-09-29 12:12:03

Chatbots are genuinely impressive when you watch them do things they're good at, like writing a basic email or creating weird, futuristic-looking images. But ask generative AI to solve one of those puzzles in the back of a newspaper, and things can quickly go off the rails. That's what researchers at the University of Colorado at Boulder found when they challenged large language models to solve sudoku. And not even the standard 9x9 puzzles. An easier 6x6 puzzle was often beyond the capabilities

Topics: ai like models puzzles solve

Shop Amazon

Microsoft favors Anthropic over OpenAI for Visual Studio Code

theverge.com Tom Warren 2025-09-29 13:08:19

is a senior editor and author of Notepad , who has been covering all things Microsoft, PC, and tech for over 20 years. Posts from this author will be added to your daily email digest and your homepage feed. Microsoft is adding automatic AI model selection to its Visual Studio Code editor that will automatically pick the best model for “optimal performance.” This new auto model feature will select between Claude Sonnet 4, GPT-5, GPT-5 mini and other models for GitHub Copilot free users, but pai

Topics: ai microsoft model models openai

Shop Amazon

Google releases VaultGemma, its first privacy-preserving LLM

arstechnica.com Unknown 2025-09-30 01:04:04

The companies seeking to build larger AI models have been increasingly stymied by a lack of high-quality training data. As tech firms scour the web for more data to feed their models, they could increasingly rely on potentially sensitive user data. A team at Google Research is exploring new techniques to make the resulting large language models (LLMs) less likely to "memorize" any of that content. LLMs have non-deterministic outputs, meaning you can't exactly predict what they'll say. While the

Topics: budget data models noise training

Shop Amazon

Crowdstrike and Meta just made evaluating AI security tools easier

zdnet.com Webb Wright 2025-09-30 17:29:27

fotograzia/Moment via Getty Images Follow ZDNET: Add us as a preferred source on Google. ZDNET's key takeaways AI is both a cybersecurity threat and a solution. Benchmarks will test LLMs for real-world cybersecurity tasks. The suite could help developers build better models. Overwhelmed with cybersecurity tool options? A new set of benchmark tests aims to help you evaluate them and find the right ones for you. Also: Navigating AI-powered cyber threats in 2025: 4 expert security tips for b

Topics: ai benchmarks cybersecurity models powered

Shop Amazon

Fixing Hallucinations Would Destroy ChatGPT, Expert Finds

futurism.com Unknown 2025-09-30 20:33:13

In a paper published earlier this month, OpenAI researchers said they'd found the reason why even the most powerful AI models still suffer from rampant "hallucinations," in which products like ChatGPT confidently make assertions that are factually false. They found that the way we evaluate the output of large language models, like the ones driving ChatGPT, means they're "optimized to be good test-takers" and that "guessing when uncertain improves test performance." In simple terms, the creator

Topics: ai hallucinations models openai users

Shop Amazon

OmniFocus 4.8 gets on-device Apple Intelligence support

9to5mac.com Marcus Mendes 2025-10-01 05:26:25

It should come as no surprise that Omni Group is ready for macOS Tahoe 26, iOS 26, iPadOS 26, and visionOS 26 on day one, with a full Liquid Glass revamp across its entire suite of apps. But beyond the visual overhaul, one of the most interesting additions is the support for Apple’s Foundation Models framework in OmniFocus 4.8. Here’s what it can do. Quick recap on Apple’s Foundation Models framework With today’s updates, Apple is introducing the Foundation Models framework, which lets develo

Topics: 26 apple automation models omni

Shop Amazon

Show HN: Semlib – Semantic Data Processing

news.ycombinator.com Unknown 2025-10-01 04:45:56

Semlib Semlib is a Python library for building data processing and data analysis pipelines that leverage the power of large language models (LLMs). Semlib provides, as building blocks, familiar functional programming primitives like map , reduce , sort , and filter , but with a twist: Semlib's implementation of these operations are programmed with natural language descriptions rather than code. Under the hood, Semlib handles complexities such as prompting, parsing, concurrency control, caching,

Topics: computation data models processing semlib

Shop Amazon

La-Proteina

news.ycombinator.com Unknown 2025-09-28 10:43:58

La-Proteina: Atomistic Protein Generation via Partially Latent Flow Matching Abstract. Recently, many generative models for de novo protein structure design have emerged. Yet, only few tackle the difficult task of directly generating fully atomistic structures jointly with the underlying amino acid sequence. This is challenging, for instance, because the model must reason over side chains that change in length during generation. We introduce La-Proteina for atomistic protein design based on a n

Topics: atom model models motif yaml

Shop Amazon

OpenAI Realizes It Made a Terrible Mistake

futurism.com Unknown 2025-10-01 16:30:56

OpenAI claims to have figured out what's driving "hallucinations," or AI models' strong tendency to make up answers that are factually incorrect. It's a major problem plaguing the entire industry, greatly undercutting the usefulness of the tech. Worse yet, experts have found that the problem is getting worse as AI models get more capable. As a result, despite incurring astronomical expenses in their deployment, frontier AI models are still prone to making inaccurate claims when faced with a pr

Topics: ai hallucinations language models openai

Shop Amazon

‘Selling coffee beans to Starbucks’ – how the AI boom could leave AI’s biggest companies behind

techcrunch.com Russell Brandom 2025-10-01 17:00:00

How much do foundation models matter? It might seem like a silly question, but it’s come up a lot in my conversations with AI startups, which are increasingly comfortable with businesses that used to be dismissed as “GPT wrappers,” or companies that build interfaces on top of existing AI models like ChatGPT. These days, startup teams are focused on customizing AI models for specific tasks and interface work, and see the foundation model as a commodity that can be swapped in and out as necessary

Topics: ai companies foundation model models

Shop Amazon

California Lawmakers Once Again Challenge Newsom’s Tech Ties with AI Bill

gizmodo.com Aj Dellinger 2025-10-01 13:15:29

Last year, California Governor Gavin Newsom vetoed a wildly popular (among the public) and wildly controversial (among tech companies) bill that would have established robust safety guidelines for the development and operation of artificial intelligence models. Now he’ll have a second shot—this time with at least part of the tech industry giving him the green light. On Saturday, California lawmakers passed Senate Bill 53, a landmark piece of legislation that would require AI companies to submit

Topics: ai california companies models safety

Shop Amazon

Which iPhone 17 Model Should You Buy?

wired.com Julian Chokkattu 2025-10-03 03:00:00

Apple’s 2025 iPhones are here, and things are quite different for the first time in a while. The base iPhone 17 will still feel familiar, but the iPhone 17 Pro models have a completely new look, and there's a brand-new model called the iPhone Air. The “Air” branding has been somewhat diluted of late—the current-gen iPad Pro models are lighter than the iPad Air—but the iPhone Air brings meaning back to the original idea: a super-thin and ultra-lightweight device. Preorders are live, and official

Topics: 17 apple iphone models new

Shop Amazon

Microsoft and OpenAI have a new deal that could clear the way for an IPO

theverge.com Richard Lawler 2025-10-04 01:00:03

is a senior editor following news across tech, culture, policy, and entertainment. He joined The Verge in 2021 after several years covering news at Engadget. Posts from this author will be added to your daily email digest and your homepage feed. As OpenAI attempts to restructure itself and eventually go public, a hurdle for the startup, recently valued at $500 billion, is its increasingly complicated partnership with Microsoft. On Thursday afternoon, the two companies released this joint state

Topics: agreement billion microsoft models openai

Shop Amazon

Microsoft is making ‘significant investments’ in training its own AI models

theverge.com Tom Warren 2025-10-04 05:21:17

is a senior editor and author of Notepad , who has been covering all things Microsoft, PC, and tech for over 20 years. Posts from this author will be added to your daily email digest and your homepage feed. Microsoft AI launched its first in-house models last month, adding to the already complicated relationship with its OpenAI partner. Now, Microsoft AI chief Mustafa Suleyman says the company is making “significant investments” in the compute capacity required to Microsoft’s own future fronti

Topics: ai microsoft models said suleyman

Shop Amazon

Apple denies Politico report on AI guideline changes around DEI, vaccines, and Trump

9to5mac.com Marcus Mendes 2025-10-04 12:41:29

Politico has published an extensive report claiming that, following Trump’s election, Apple changed its AI training guidelines on issues such as DEI, vaccines, elections, and Trump himself. Here are the details. Data annotation It is common practice for tech companies to rely on subcontractors to help with the labeling and post-training process of their AI models. Politico’s report says that Apple contracts Transperfect, a company that offers “translation services and solutions,” including AI

Topics: ai apple models politico report

Shop Amazon

Latest Tech News

Is AI Capable of 'Scheming?' What OpenAI Found When Testing for Tricky Behavior

‘AI Scheming’: OpenAI Digs Into Why Chatbots Will Intentionally Lie and Deceive Humans

How developers are using Apple’s local AI models with iOS 26

Rupert's snub cube and other Math Holes

OpenAI’s research on AI models deliberately lying is wild

Luma AI's New Ray3 Video Generator Can 'Think' Before Creating

Boring is good

White House officials reportedly frustrated by Anthropic’s law enforcement AI limits

Irregular raises $80 million to secure frontier AI models

AI models know when they're being tested - and change their behavior, research shows

Ask HN: What's a good 3D Printer for sub $1000?

With Lumo, Proton thinks it can carve a place at the AI table

Trump Mobile’s earliest actual phones are neither made in America, nor cheap

How Google's new AI model protects user privacy without sacrificing performance

Microsoft Favors Anthropic over OpenAI for Visual Studio Code

AI Is Bad at Sudoku. It's Even Worse at Showing Its Work

Microsoft favors Anthropic over OpenAI for Visual Studio Code

Google releases VaultGemma, its first privacy-preserving LLM

Crowdstrike and Meta just made evaluating AI security tools easier

Fixing Hallucinations Would Destroy ChatGPT, Expert Finds

OmniFocus 4.8 gets on-device Apple Intelligence support

Show HN: Semlib – Semantic Data Processing

La-Proteina

OpenAI Realizes It Made a Terrible Mistake

‘Selling coffee beans to Starbucks’ – how the AI boom could leave AI’s biggest companies behind

California Lawmakers Once Again Challenge Newsom’s Tech Ties with AI Bill

Which iPhone 17 Model Should You Buy?

Microsoft and OpenAI have a new deal that could clear the way for an IPO

Microsoft is making ‘significant investments’ in training its own AI models

Apple denies Politico report on AI guideline changes around DEI, vaccines, and Trump

About GoKawiil

Privacy

Advertising

Latest Tech News

Trending Topics

Hot Now

Popular

Emerging

About GoKawiil

Privacy

Advertising