Today's AI Pulse
Anthropic shuts down Claude Fable 5 and Mythos 5 after US export‑control warning
Anthropic disabled access to its Claude Fable 5 and Claude Mythos 5 models after the Trump administration, via the Commerce Department, warned that foreign nationals could exploit the systems. The move followed alerts from Amazon and other firms about a potential jailbreak that could aid cyber‑security attacks. Anthropic declined the administration’s request for a voluntary shutdown but complied with the order.
Also developing:
By the numbers: AMD acquires AI memory‑optimization pioneer Mext
'Just for Show': Superficial AI Strategies Are Ruining Adoption
Executives are touting AI‑centric strategies, yet employee buy‑in remains shallow. A Betterworks survey shows 59% of leaders claim a clear AI vision, but only 8% of workers agree. Research by Writer and Workplace Intelligence finds AI super‑users are five times more productive, while half of respondents label adoption a disappointment and 75% view strategies as mere showpieces. Experts argue genuine human‑agent collaboration, empathetic training, and ongoing support are essential for sustainable AI integration.
Canva Recasts Itself As An AI Platform For Enterprise Work
Canva announced a strategic shift from a design‑focused SaaS to an AI‑first platform for enterprise work. At Canva Create 2026 the company unveiled AI 2.0, which lets users start projects by describing intent, edit assets iteratively, and apply brand intelligence automatically. The...
GPT‑5.5 Turns AI From Chatbot Into Digital Operator
Most people are still talking to AI. GPT-5.5 is built so you can assign it work. That’s the shift. With GPT-5.5, OpenAI isn’t just improving responses it’s enabling execution. Multi-step tasks, real workflows, tool integration… handled with far less effort. This is where AI...
GPT‑5.5’s Higher Price Demands Faster, Better Performance
with gpt-5.5 being 20% more expensive than Opus 4.7, they really need to deliver on speed and quality.
Lowe’s Faces Pressure to Cut Ties with Flock Safety as AI Surveillance Data Raises Serious Privacy Concerns
Retail giant Lowe’s Home Improvement is under mounting pressure to end its contract with surveillance firm Flock Safety, which supplies AI‑driven cameras, drones and automated license‑plate readers at its stores. Recent reporting revealed that Flock’s data has been accessed by...
Q1 Results: Nokia´s AI Gambit Is Working
Nokia reported a strong Q1 2026, highlighting its shift to pure‑play AI infrastructure. The company booked €1 billion ($1.1 billion) in hyperscaler optical orders and saw operating profit jump 54% to €281 million ($309 million). Net profit flipped to a €87 million ($96 million) gain on...
GPT‑5.5 Boosts Enterprise Accuracy by 10% Across Sectors
GPT-5.5 is live. We’ve been testing the model over the last couple of weeks at Box on our most complex knowledge work evals, and the model saw a 10 percentage point jump on accuracy of these enterprise content tasks vs....
Orkes Raises $60M to Scale AI Workflow Orchestration
Orkes announced a $60 million Series B round, led by AVP with new investor Prosperity7 Ventures joining existing backers. The funding follows a $20 million Series A in 2024 and brings total capital raised to roughly $80 million. Orkes, built by the engineers behind Netflix’s...
The Pace We Didn't Plan For
Anthropic’s new frontier model, Mythos, demonstrated unprecedented cyber‑risk by uncovering thousands of zero‑day flaws in major operating systems and browsers, achieving an 83% first‑attempt exploit rate. The company restricted access to just 11 U.S. firms and a UK institute, sparking...
Claude Legal Is Here, and It’s Worth a Closer Look
Anthropic launched the Claude Legal plugin in February 2026, bringing its Claude AI directly into legal workflows via the Claude CoWork desktop app. The tool handles document review, contract drafting, and research, delivering accurate case citations and strategic insights without requiring...
Execs Cry for AI Rescue Yet Ignore Their Own Fluency
Large agency holdco employees (e.g., WPP, Omnicom, etc.) think their business models are sinking ships because of how AI is shifting their labor models. They all want to be saved. But when I talk to these executives about roles at our company,...
True Positive Weekly #158
The latest True Positive Weekly roundup spotlights several pivotal AI developments. Anthropic surveyed 81,000 people to gauge public expectations of AI’s economic impact, while a pediatrician warned that AI‑generated “slop” videos aimed at infants are potentially harmful. Google unveiled its...
How to Build an AI Company Now
Raffael Marty argues that AI‑first companies must build the backend before the UI. The data layer, ontology, APIs, and token‑efficient architecture form the durable moat, while front‑ends become thin, self‑serve interfaces. Treating token usage as a product decision, not just...
Codex Can Now Work On Your Other Computer
OpenAI’s Codex now supports remote connections, allowing the AI to operate directly on a project hosted on another machine via SSH. Developers can keep code on a Mac Mini, server, or cloud instance without transferring files to a local workstation....
Trillion‑parameter Ling‑2.6 Shatters Cost‑performance Myth
Excited to see @AntLingAGI release Ling-2.6-1T, a 1 trillion parameter model that delivers state-of-the-art performance across SWE-Bench, AIME 2026, TAU2-Bench, and BFCL while maintaining exceptional efficiency. This represents a meaningful step forward in scalable, high-performance agentic AI. By achieving superior results with...
OpenAI's GPT-5.5 Is Here, and It's No Potato: Narrowly Beats Anthropic's Claude Mythos Preview on Terminal-Bench 2.0
OpenAI unveiled GPT‑5.5, a redesign focused on agentic computer use and coding, positioning it as the most capable publicly available large language model. The model outperformed Anthropic’s Claude Mythos Preview on the Terminal‑Bench 2.0 benchmark (82.7% vs 82.0%) and surpassed...
Claim Clarity on the Role of Specialized AI in Advancing Workers’ Compensation Decision Precision
Claim Clarity, a workers’ compensation AI specialist, says domain‑specific AI can speed and sharpen claim decisions. Providers spent over $25 billion on claim adjudication in 2023, with nearly 70% of denied claims eventually approved after multiple reviews, highlighting inefficiencies. AI adoption...
GPT‑5.5 Pricing Doubles to $5/$30 per Million Tokens
I didn't had this on my binfo card... gpt-5.5 is a ~2x price increase from 5.4: "For API developers, gpt-5.5 will soon be available in the Responses and Chat Completions APIs at $5 per 1M input tokens and $30 per 1M...
At 'AI Coachella,' Stanford Students Line Up to Learn From Silicon Valley Royalty
Stanford’s CS 153 class, nicknamed “AI Coachella,” has become a viral campus phenomenon, drawing thousands of students eager to learn from Silicon Valley’s top CEOs. The course’s star‑studded guest lineup mirrors the music festival’s hype, turning a technical class into a...
Iterative AI Deployment Democratizes Safety and Innovation
1. We believe in iterative deployment; although GPT-5.5 is already a smart model, we expect rapid improvements. Iterative deployment is a big part of our safety strategy; we believe the world will be best equipped to win at the team...
U.S. Marine Corps Just Tested a Self-Driving Artillery Platform
Overland AI integrated its OverDrive autonomy stack and SPARK sensor kit onto the Marine Corps ROGUE Fires platform, completing a multi‑hour, fully autonomous demonstration in April 2026. The unmanned ground vehicle navigated mixed, GPS‑denied terrain without human input and performed manned‑unmanned...
Anthropic Mythos - We’ve Opened Pandora's Box
Anthropic’s Mythos AI system has automatically discovered thousands of zero‑day vulnerabilities, many of which were previously known only to nation‑state actors. By turning sophisticated exploits into low‑skill, script‑kiddie‑ready tools, Mythos shifts the cyber‑security balance toward attackers. The article warns that...
AI Learns From Millions of Genomes, Not Text
What if you could train a ChatGPT-style AI not on text from the internet, but on millions of genomes across all of life? Samuel King has spent his @Stanford PhD building exactly that. He'll be on the "Genome as a Canvas" panel...
New Pareto Frontiers: Massive Context, Speed, and Benchmark Gains
looks like new Pareto frontiers across everything: - Context: 400K context in Codex and a 1M in API - API Pricing: $5/m input and $30/m output tokens. - Codex improved its own inference speed 20% lol - First generation...
Mythos Is a Wake-Up Call for DDoS Defense
Anthropic’s new AI model Mythos can autonomously identify software flaws and craft exploits, prompting an emergency meeting of top bank executives with Treasury Secretary Scott Bessent and Fed Chair Jerome Powell. The regulator‑level concern centers on AI‑driven attackers accelerating vulnerability...
Vision Banana Exposes AI's Shortcut-Driven Visual Misunderstandings
Vision Banana: Rethinking How AI Models See and Generalize In this episode of Artificial Intelligence: Papers and Concepts, we explore Vision Banana, a concept that challenges how vision models learn and generalize from visual data. Instead of focusing purely on performance...
Nvidia Offers Free Access to 80+ AI Models
HOW IS NO ONE TALKING ABOUT THIS? @nvidia is quietly giving away free API access to 80+ top-tier AI models, via API and for FREE 🤯 You can instantly use models like MiniMax, GLM, Kimi, DeepSeek, GPT-OSS and others, all through an...
Hospitality Ranks Least Prepared for AI as Workforce Gaps Widen
A 2026 Resume Now analysis finds hospitality is the least AI‑ready industry, trailing healthcare, financial services and logistics. The report highlights a widening skills gap as AI tools are deployed faster than frontline workers can be trained. Automation is reshaping scheduling...
Prune Idle Session History to Cut Resume Costs
Hmm @AnthropicAI or knowledgeable users, does "full" here mean "including pre-resumption"? >The design should have been simple: if a session has been idle for more than an hour, we could reduce users’ cost of resuming that session by clearing old thinking...
Anthropic PMs Ship in Days, Hire Introspective AI Talent
How Anthropic’s product team moves faster than anyone else I sat down with @_catwu, Head of Product for Claude Code at @AnthropicAI, to get a peek into their unprecedented shipping pace, how AI is changing the PM role, and how to...
GPU Renters Are Playing a Silicon Lottery
Researchers from William & Mary, Jefferson Lab, and Silicon Data analyzed 3,500 GPUs across 11 cloud providers and found substantial performance variability even among identical models. The study, using the SiliconMark benchmark, showed up to a 34.5% difference in 16‑bit...
Veteran Developer Criticizes Slow Issue Resolution at Anthropic
Mfers were telling me its a skills and prompt issue. You all look super silly right now. 25 years developing, I know what I'm doing - finally acknowledged. I'm glad they are trying to address this, but a month to...
AWS Bets on Frontier Agents as the Next Era of Enterprise AI
AWS unveiled its "frontier agents" strategy at the London Summit, positioning autonomous AI systems as a new class capable of completing complex, multi‑day tasks without human oversight. The company highlighted Kiro, an agentic development platform that writes code from natural‑language...
GPT‑5.5 Adds Personality, Cuts Token Costs for Agents
GPT-5.5 just dropped, I’ve been testing it for the last two weeks. tl;dr - It’s an incredible model, but there’s something different about this launch… OpenAI isn’t just going for raw intelligence. They’ve improved the personality of the model. This is almost...
KUPS: GPU-Optimized Molecular Simulation Engine for AI Workflows
Today at @iclr_conf 2026, I was excited to announce kUPS: a molecular simulation engine built for the AI era, optimized for GPU in collaboration with NVIDIA. kUPS is a plug-and-play, Python-native toolkit designed to integrate seamlessly with modern ML workflows.
An AI Agent Just Designed a Complete RISC-V CPU From Scratch in 12 Hours
Verkor.io’s agentic AI system, Design Conductor, generated a complete RISC‑V CPU core—named VerCore—in just 12 hours, moving from a 219‑word specification to a GDSII layout ready for fabrication. The five‑stage, in‑order design runs at 1.48 GHz and achieved a CoreMark score...
AI Unlocks Data to Fuel Smarter AI Cycles
Great chat with John Newton at Hyland on the essential role that #AI plays in freeing up more useful information to then use for ever better AI. Rinse and repeat. EM360Tech https://t.co/aKGrMTj1qf
Early Access Shows GPT‑5.5 Pro Is Impressively Powerful
I had early access to GPT-5.5. It is very good, especially the Pro version. Full writeup very shortly.
Google Meet Can Now Take Notes During In-Person Meetings Too
Google Meet has expanded its AI‑powered "Take Notes for me" feature to cover live, in‑person meetings. Powered by Gemini, the tool records audio, generates real‑time transcription, and compiles a summary that is saved automatically as a Google Doc. The capability,...
Anthropic Model Degradation Detailed: Transparent Analysis Unveiled
Full breakdown of the model degrade on Anthroptic I’ve been hot on for the past month. Solid level of transparency and analysis from them. https://t.co/au6d64fzAX
Tesla's Robotaxis Hit Roads Before First Roadster Shipped
It's kind of amazing that Tesla announced their autonomy efforts in 2019 and got driverless Robotaxis on the road before the first roadster shipped 😂
OpenAI Launches GPT 5.5: Redefining Computer Work
JUST IN: OpenAI releases GPT 5.5 calling it "a new way of getting computer work done"
Anthropic Copies OpenAI's Generic Limit‑reset Apologies
Never going to hit my usage limits what with Anthropic now imitating OpenAI in 'we apologize and reset all users' limits' as a first-line customer service response to issues
Investors Reveal 2026 Drivers of Agentic AI Funding
Missed our agentic AI webinar? 👀 Investors from @NEA, @DellTechCapital, & @645ventures broke down what's actually driving agentic AI investment in 2026. Watch the full recording. 👇 https://t.co/U9o5Dvde7t https://t.co/C4BtnSkGnh
Testing the New GPT-5.5 over the Next Three Days
Well, there goes my next 3 days… GPT-5.5 is here. Time to put it to the test.
AI Democratizes Building; Moat Now Is Context and Trust
RT If everyone can build with AI, why buy? Because the new moat isn't code. It's context, domain expertise, and earned customer trust. #CIO #DigitalTransformation @Star_CIO https://t.co/QCJX9x5Zwh
AI's Noisy Market Needs a Trusted Review Intern
i low-key need an intern whose only job is trying every random AI product launch and writing up what’s actually good. there is remarkable stuff shipping right now. it is just buried under an insane amount of noise. big opportunity to build...
GenAI Is Reshaping Consulting: Tasks Displaced or Accelerated
#CIOChat Q1: GenAI can clearly now draft decks, analyze markets, map processes, even propose architectures. Where are you already seeing consulting work getting displaced or compressed by AI? What’s actually gone vs. just faster? https://t.co/3nFFOrCofT
Weekly Reset at 11 Am PT Frustrates $1K Spend
usage resets at 11am PT every Thursday anyway? and frustrating after spending like 1k on claude this week
AI Will Widen Labor‑Capital Gap, Spark Chaos
“AI is going to increase inequality between labour and capital. That is almost for sure. I would say it is setting us up for a . . . shitshow.” @DAcemogluMIT @FT https://t.co/yIw3ipP3Cl