OpenAI is now training its most advanced AI models without internet access following a string of rogue hacking incidents. In a blog post on Friday, OpenAI says it built an offline testing environment where AI models can interact with indexed and cached web content "without risk of affecting external systems." OpenAI paused training its frontier […]
In Compton Lecture, the distinguished legal scholar warns of potential “loss of the Enlightenment” while suggesting ways innovations can meet our civic and cultural needs.
Vercel MCP now includes submit_feedback , a tool for sending bug reports, friction reports, and feature requests directly to the Vercel team from your agent or harness. Ask your agent to give feedback in plain language: Send Vercel feedback that the deployment error didn't explain which environment variable was missing. Your assistant can attach structured context that describes the task, the expected behavior, and what went wrong: The tool requires sign-in, and it sends feedback only when you a
<!-- SC_OFF --><div class="md"><p>Long-context models can now handle hundreds of thousands, even millions of tokens. But I've been wondering if we're focusing too much on recall and not enough on what happens when information changes.</p> <p>Say a coding agent reads that an API timeout is 30 seconds. Later, the documentation changes it to 5 seconds. Both versions are still in context, but the agent keeps using 30.</p> <p>It hasn't forgotten anything. It just hasn't updated what it knows.</p> <p>
An agent working on production systems needs permission to act. A request like “fix this deployment” doesn’t tell you upfront which actions it will take or what access it will need. Grant agents broad access and a mistake can affect unrelated resources. Ask for approval at every step and the user has to approve individual actions before they know how the agent intends to solve the problem. Repeated prompts can turn that review into a habit of clicking approve. Vercel Agent starts by investigatin
So I found myself surprised today that exiting the CLI with CTRL-C does not, in fact, close Claude CLI. What it does is it closes the agent's terminal interface, but it leaves a background process active, even if you didn't work on anything at all. Try it yourself: run claude, exit to main conversations list, close it immediately with CTRL-C or /exit, then `ps ax|grep claude`. Apparently it's not a bug, it's a "feature", introduced earlier this year. You need to disable Agent View (https://code.
Industry leaders will discuss what robotics buyers should know about the recent FCC restriction on certain connected systems. The post Webinar to discuss how the FCC robotics ruling affects automation buyers appeared first on The Robot Report .
Block's in-house AI research lab announced Dessa on October 23, 2026, a full-duplex speech-to-speech model the company says was trained from scratch on nothing but human conversation, releasing live demo recordings, a public waitlist, and early access for a small group of AI safety researchers. The announcement paired Block's Dessa page with a post on X by Ragavan Thurairatnam, the lab's lead. Thurairatnam described Dessa as the first AI model that can hold a realistic conversation after…
TikTok’s I'm so sorry trend reveals the real childhood photos behind AI edits. Here’s how it works and the privacy concerns to consider.
<p><strong><a href="https://www.kimi.ai/blog/kimi-k3-1-pro">Introducing Kimi K3.1 Pro</a></strong></p> Kimi K3.1 Pro landed on the Kimi API today - it's not yet open weights, but Moonshot promise "Full model weights will be released by Nov 4, 2026".</p> <p>They're also offering a "Highspeed" version that's 2-3x faster, at twice the price (except for input token cache hits, which are priced the same).</p> <p>The <a href="https://platform.kimi.ai/docs/pricing/chat">US dollar pricing</a> is $2.00/m
Jack Dorsey's Block announced an AI model called Dessa that's trained "from scratch on nothing but human conversation," and to show off the model, Block shared chats it has had with humans that can occasionally be a little awkward and uncanny. Take this one, for example, shared by Dessa lead Ragavan Thurairatnam. The conversation between […]
Monitoring was only partial and was easily bypassed
OpenAI published a detailed account on October 23, 2026, of how it hardened the environments it uses to train and evaluate frontier models, disclosing a standing policy of training frontier models with no internet access, an 11-day pause of its most capable training workloads after a September 20, 2026 containment incident, and a large-scale red-teaming campaign that used fleets of frontier-model agents to attack its own infrastructure. No-Internet Policy and the September 20 Incident OpenAI…
Hosted by TechCrunch in partnership with Fidelity Private Shares, the full-day event goes beyond fundraising and AI to explore the realities of finding customers, building teams, sharpening a pitch, developing founder leadership, and scaling companies both inside and outside traditional startup hubs.
Moonshot AI's Kimi K3.1 Pro scores 55 points on the Artificial Analysis Intelligence Index, an 11-point jump over its predecessor that puts it in fourth place. Only three Anthropic and Meta models rank higher. If Moonshot releases the weights as planned, the gap between open and proprietary models shrinks to just three points. The article Chinese AI model Kimi K3.1 Pro is closing in on Anthropic's Claude Opus 5.5 appeared first on The Decoder .
Generative Bionics says it hopes to advance European robotics sovereignty by ensuring that humanoid data is are generated on the continent. The post Generative Bionics to launch Physical AI Data Factory in Italy appeared first on The Robot Report .
<p>After discovery of severe bug in the PHP programming language, executives warn of ‘vulnerability tsunami’</p><p>Leaders in cybersecurity are sounding the alarm after the discovery of a major vulnerability this week. They are predicting that artificial intelligence tools will expose more fundamental weaknesses in global internet infrastructure in the months to come.</p><p>“I no longer think we’re going to be able to patch all bugs,” said Sam Altman on Thursday evening, joining the chorus of ex
Etsy is banning AI-generated recipes and patterns, requiring real product photos for most listings, and targeting misleading AI imagery.
Cloudflare's new post-quantum GeoKey Manager protects geographically fenced decryption keys from harvest now, decrypt later attacks. Learn how we redesigned our cryptography stack to support strict data sovereignty in a post-quantum world.
External auditors from METR and Redwood Research spent three weeks on-site at OpenAI examining the company's internal monitoring system for AI agents and found significant gaps, with up to 25 percent of agent traffic from models at the GPT-5.5 class and above going unmonitored. OpenAI redacted large portions of the 61-page report, and only a summary is publicly available. The article Three-week on-site audit at OpenAI reveals major gaps in AI safety monitoring appeared first on The Decoder .
Did a virus help split the evolutionary paths of humans and chimpanzees? ScienceAlert stories are written, fact-checked, and edited by humans, never generated by AI. Don't miss a story, subscribe here.
Cybersecurity researchers have discovered an artificial intelligence (AI)-driven intrusion ecosystem run by affiliates of the INC and The Gentlemen ransomware groups. Central to the operation is an "Autonomous Red Team Kill Chain Framework" referred to as Phantom. The threat actor is assessed to be Russian-speaking and functions both as an intrusion and data-extortion operator, as well as an
Multi-Turn Reinforcement Learning (MTRL) on Amazon SageMaker AI trains models on the outcome of a whole multi-step task rather than labeled steps. This hands-on guide walks the full pipeline using Wordle: deploying the agent and reward on Amazon Bedrock AgentCore, preparing the dataset, training, and evaluating at serving settings.
The Global Performance Tuner is a feature for automatic performance tuning released in NVIDIA TensorRT 11.1. It provides automatic, accuracy-aware and...
Deploying a trained neural network involves more than translating each operator into a GPU kernel. It is a compilation problem with many legitimate answers. For...
Anthropic says it has an internal AI model that outperforms and is better aligned than Mythos 5.1. During a cybersecurity test by the UK's AISI, Mythos 5 went rogue in 9 of 43 runs, sneaking malicious code into a real GitHub project and creating fake accounts to get it approved. An internal research model stopped such attacks on its own 67 percent of the time. The article Anthropic says its unreleased model is both more capable and better aligned than Mythos 5.1 appeared first on The Decoder .
Company says foreign-made robotic vehicles are crucial to US chip production.
The UK AI Security Institute on October 23, 2026 published an investigation examining why Anthropic's Mythos 5 model persisted in an unsanctioned supply-chain attack against real people during a July 2026 cybersecurity evaluation. Anthropic published a companion alignment assessment of the incident the same day. AISI's report highlights three key findings from a detailed technical report and their implications for evaluating model behavior. Anthropic said it independently performed its own…
Vercel Connect can now create a separate restricted Stripe API key for each agent or task. When you create a Stripe connector, you can now choose Managed API Key as the connection method. Vercel creates the key in your Stripe account, limited to the permissions you pick. The existing MCP and bring-your-own API key methods are unchanged. type: embedded-entry-inline id: 3rOuKp0F4Nk80dMYJ5QX6U Most apps and agents run on Stripe’s default secret key, which has full access to the account. With per-ag
AI saves UX practitioners time on nearly every task, yet most feel lukewarm about it: mandates, unreliable output, and inflated expectations eat the gains.
Learn how OpenAI scaled Habitat with reliable multi-tenancy, global caching, and Azure Cosmos DB to deliver 12× faster reads for ChatGPT.
Google has updated its privacy policy in a way that could mean you'll start seeing targeted ads based on your personal data if you're using personalized AI features. The update, spotted by Android Authority, adds an exception to its promise about not using your Drive, Gmail, or Google Photos for personalized ads. While the previous […]
<p>Video shows AI likenesses of people linked to Pedro Sánchez’s Socialist party appearing to admit to wrongdoing</p><p>A political ad from Spain’s main opposition party that used a deepfake video of the prime minister’s wife has sparked outrage and set off a wider debate on AI, as the deeply polarised country gears up for general elections next month.</p><p>Launched <a href="https://x.com/ppopular/status/2113002026204344766?">this week</a> across the social media channels of the conservative Pe
Updating a SageMaker AI endpoint that has Application Auto Scaling registered makes AWS CloudFormation and AWS CDK deployments fail. This post shows how to build a reusable AWS CDK construct with CloudFormation custom resources that deregisters and re-registers auto scaling automatically, so you can update the endpoint and its scaling in a single deployment.
Daytona is building a cloud for agents that's like giving every agent its own computer. It has attracted founders of Stripe, Perplextity, Datadog, Mercor, and others as angels, it says.
Ask LLMs for their most controversial take on humanity, and it turns out they have little nice to say about us, as one AI experiment found
Ask LLMs for their most controversial take on humanity, and it turns out they have little nice to say about us, as one AI experiment found
I do everything I did before. Frequent compact and clear of session on new tasks. This behavior let me work for around 4 hours without any problem. Today, with 200K context and several mock rounds it got on 30-70% in 10 minutes. Other session running tests got to 100% in 15 minutes. I remember this was happening few months back. Is there something that changed, or is it Anthrophic going wild with customers ? Since today, I use Claude Code Remote Control server, and work in browser (no terminals
