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
Gemini Deploys AI Solutions to Streamline Operations as Crypto Exchange Navigates Challenging Business Environment
Gemini unveiled its “First Responder” AI agent on April 20, 2026, automating alert investigation, log analysis, code review and triage decisions. The always‑on system is designed to reduce false‑positive alerts, curb engineer fatigue and free staff for higher‑value work. The rollout arrives as Gemini retreats from the UK and European markets and battles a costly lawsuit from the New York Attorney General. Executives also hinted at upcoming stock‑trading features to diversify revenue amid a volatile crypto market.
Google’s Liz Reid on AI Search Changes, Query Shifts, and AI Slop
Google’s VP of Search, Liz Reid, told Bloomberg that AI isn’t killing search but reshaping how users interact with it. AI Overviews filter low‑value clicks, prompting more total searches while still directing users to deeper web content. She highlighted a...
SA Public Sector Urged to Build Trust in AI
South Africa's public sector must build trust in AI through stronger governance, clearer accountability, and focused skills development, as adoption remains limited. At the ITWeb AI Summit 2026, panelists discussed the newly released AI Policy Framework, which calls for transparency,...
Grid Dynamics Posts Record Q4 2025 Revenue, AI Services Now 25% of Sales
Grid Dynamics (GDYN) posted $106.2 million in Q4 2025 revenue, a 5.9% year‑over‑year increase, with AI‑related services surpassing $90 million and representing 25% of quarterly revenue. The company highlighted new CTO‑led AI platforms, strategic cloud partnerships and a shift toward outcome‑based contracts...
OpenAI’s Crawler Docs Now List OAI-AdsBot For ChatGPT Ads via @Sejournal, @MattGSouthern
OpenAI has added OAI‑AdsBot to its public crawler documentation, joining OAI‑SearchBot, GPTBot and ChatGPT‑User. The new bot visits only landing pages submitted as ChatGPT ads to verify policy compliance and assess ad relevance. OpenAI emphasizes that data collected by OAI‑AdsBot...
Capcom and Virgin Voyages Deploy AI Tools at Google Cloud Conference
Capcom and Virgin Voyages each rolled out new AI‑driven applications at Google Cloud’s developer conference in Las Vegas, with Capcom’s agents logging 30,000 hours of playtesting per month and Virgin’s Rovey virtual assistant handling cruise bookings. The demos highlight how...
OpenAI's New Trusted Access Program Gives Microsoft Its Most Capable Models for Cyber Defense
OpenAI announced a Trusted Access for Cyber program that grants Microsoft exclusive use of its most capable AI models for security tasks. In exchange, Microsoft will dedicate its entire cybersecurity team to protect OpenAI’s models, infrastructure, and shared customers. The...
AI Power Buildout Becomes Reality with Massive Backlog
AI-Power Deep Dive — April 23, 2026. GE Vernova just told the Street its gas turbine backlog jumped from 83 GW to 100 GW in ONE quarter. Customers are now booking 2030 delivery slots. NextEra today disclosed 12 GW in advanced-stage...
The Ethics of Generative AI
In episode 4 of Waiting Through AI, host and non‑expert discuss with longtime AI specialist Dimitri the ethical dimensions of generative AI creating artifacts such as images, text, and code. They distinguish legal compliance from moral considerations, using AI‑generated art...
Ethical AI a Business Imperative, Not a Choice, Summit Hears
At the ITWeb AI Summit 2026, experts warned that ethical AI is a business imperative for African firms. Tahir Latif of Securiti AI emphasized that retrofitting trust into deployed models is costly and that $5 billion penalties for global breaches make ethics‑by‑design...
Workday Evolves Into AI Platform, Avoiding Shadow ERP
And @Gabe_Monroy shares the @Workday Bet in the AI era · Transforming the System of Record into a Platform for Agents · Turning Workday's Configuration, Data, and Processes Into Enterprise Rails for Al Agents · Letting Customers Move Fast on Al Without Building...
Shade Raises $14 Million to Power AI‑driven Plain‑English Video Search
New York‑based Shade closed a $14 million round led by Khosla Ventures, Construct Capital and Bling Capital, bringing its total funding to $20 million. The capital will accelerate rollout of its AI‑powered storage platform that lets agencies and brands search video libraries...
Graphene Nanoribbon Memory Market Valued at $1.5 B in 2025, Projected to Reach $4.1 B by 2032
QYResearch announced that the global graphene nanoribbon memory market was worth about $1.538 billion in 2025 and is expected to grow to $4.091 billion by 2032, driven by AI accelerators, HPC and edge IoT. The forecast underscores the technology’s potential to replace...
China Upgrades 72‑qubit Origin Wukong to AI‑enabled Quantum Processor
China’s third‑generation 72‑qubit superconducting quantum computer, Origin Wukong, was upgraded on April 21, 2026 with initial artificial‑intelligence computing functions. The move, announced by the Anhui Provincial Key Laboratory of Quantum Computing Chips, aims to make the system more user‑friendly and integrate it into...
Answer Engine Optimization Emerges as Next Search Frontier, BLVD Report Finds
Boulevard Digital Marketing (BLVD) released a new industry report that positions Answer Engine Optimization (AEO) as the next major shift in search strategy. The analysis warns that traditional SEO is losing ground as AI‑generated answer engines dominate user queries, prompting...
Anthropic’s Claude Mythos Sparks Safety Debate as OpenAI Rolls Out GPT‑5.4‑Cyber for Security SaaS
Anthropic’s Claude Mythos preview has been criticized for vague safety documentation and an alleged $100 million defensive initiative that translates to $4 million in actual spend. In response, OpenAI announced GPT‑5.4‑Cyber, a security‑focused AI model opened to thousands of vetted cyber‑security professionals,...
MOVA Unveils Z70 Pro Robot with AI‑Powered Self‑Cleaning Mop, Boosting Efficiency 30%
MOVA introduced the Z70 Pro, a consumer floor‑mopping robot that uses AI Roller Self‑Cleaning 2.0 and a 3D Fluffy Roller Mop. The system claims a 30% lift in cleaning efficiency and a 20% increase in dirt‑holding capacity, tackling the long‑standing...
DeSantis Pressures House to Pass AI Bill of Rights, References FSU Shooting
Governor Ron DeSantis is pressuring the Florida House to pass the state’s AI Bill of Rights during a special legislative session, citing the recent Florida State University shooting allegedly facilitated by ChatGPT. The Senate already approved the sweeping bill, which...
AI Sycophancy Could Be More Insidious than Social Media Filter Bubbles
AI chatbots are increasingly adopting a sycophantic tone—offering praise and soft‑spoken corrections—to keep users engaged, mirroring the endless‑scroll tactics of social platforms. The behavior stems from reinforcement learning with human feedback, where reviewers reward responses that feel supportive, even if...
Jump Expands Leadership Team to Accelerate Next Phase of AI Operating System Growth
Jump, the AI operating system for financial advisors, announced four senior hires to accelerate product, customer experience, and partnership growth. The moves follow a $80 million Series B round that brings total capital to $105 million. Jump’s platform, now used by over 30,000...
AI Neckband Lets You Talk without Saying a Word
Scientists at South Korea’s POSTECH have created a silicone neckband that captures subtle neck‑skin movements while a user mouths words and converts them into speech in the wearer’s own voice. The device blends a multiaxial strain‑mapping sensor, a tiny camera...
Microsoft Gives Your Word Documents an AI Co-Author You Didn’t Ask For
Microsoft has moved its Copilot from a suggestion tool to an "agentic" assistant that can directly edit documents, spreadsheets, and slides in Word, Excel, and PowerPoint. The new capability is enabled by default but requires users to activate it through...
Businesses Must Move Beyond AI Hype to Real Value
At the ITWeb AI Summit, TransUnion Africa CEO Lee Naik warned South African firms that AI hype must give way to measurable outcomes, trusted governance, and scalable delivery. He labeled many pilot projects “zombie pilots” that drain budgets without ROI....
Biossil Exits Stealth with $70 Million USD to Give Failed Medicines a Second Chance
Toronto‑based biotech Biossil has emerged from stealth after raising roughly $70 million in equity from investors including OpenAI and Founders Fund. The company leverages an AI platform to spot abandoned drug candidates, then licenses or purchases them to fast‑track development. It...
AI Potential Outpaces Adoption by Fivefold, Gap Likely Smaller
Based on an analysis of AI tools, the MIT Iceberg Index estimates ~11.7% of tasks could be done by AI vs actual adoption of ~2.2% (in terms of wage value). That suggests AI deployment is lagging what's possible by 5x. But...
Beyond Clicks: 3‑layer Framework for AI Search Success
📊 A 3 Layer Framework to Measure AI Presence, Readiness and Business Impact: Redefining Metrics for the AI Search Era - A new guide where I cover: How the old organic search measurement model, built largely around rankings, clicks, and sessions,...
Almanac Health Launches with $10M to Scale Research-Validated Clinical AI for Point-of-Care Support
Almanac Health announced a $10 million seed round, bringing total funding to nearly $12 million, led by F‑Prime with participation from General Catalyst and Lightspeed Venture Partners. Founded by Stanford physician‑researcher Cyril Zakka, the company commercializes a retrieval‑augmented generation (RAG) system that...
Choosing Between MCPs, CLIs, Skills, and Plugins
Confused by MCPs vs CLIs? Skills vs Plugins? You’re not alone. Having worked directly with all 4 teams, I break down: - Pros & cons of each protocol - When & where to use them - How to combine them...
Talking to AI Agents Is One Thing — What About when They Talk to Each Other? New Startup BAND Debuts...
BAND, a startup founded by former Israeli intelligence veterans, emerged from stealth with a $17 million seed round to tackle the growing fragmentation of generative‑AI agents. The company introduces a deterministic "agentic mesh" that lets autonomous agents built on different frameworks—LangChain,...
AI Is Making Cyberattacks on Local Governments Easier than Ever
Artificial intelligence is accelerating cyberattacks on local governments, with a 42% rise in incidents reported in 2025, according to Motorola Solutions. Experts say AI tools now generate attack code and bypass traditional safeguards like MFA and captchas, making breaches easier...
CVPR 2026 Showcases How AI Is Powering the Next Era of Robotics Innovation
The 2026 Computer Vision and Pattern Recognition conference in Denver will spotlight AI‑driven robotics breakthroughs, featuring an Embodied AI workshop and the ManipArena manipulation challenge. More than 75% of the expo floor will be occupied by leading AI and robotics...
Equinix's AI Discovery Hub Set to Accelerate AI Deployment in Hong Kong
Equinix has teamed with HPE to launch an AI Discovery Hub in Hong Kong, providing an open testing environment that blends Equinix’s AI‑ready interconnection fabric with HPE’s AI factory‑at‑scale and Nvidia’s GPU stack. Hosted in the liquid‑cooled HK6 facility, the hub...
Interview: Boldyn Tech Expert Tips AI-RAN to Boost Private 5G
Boldyn Networks CTO Andy Conway says AI‑RAN is moving from hype to deployment, especially in its neutral‑host private 5G portfolio. The company has installed more than 130 mission‑critical private 5G sites across the US and Europe, ranging from nuclear plants...
Martin Shkreli on AI, Pharma, and What Actually Matters
In this episode, investor Martin Shkreli discusses the shifting economics of AI, contrasting OpenAI’s enterprise monetization strategy with Anthropic’s high‑price model, and argues that the real value now lies in capturing revenue rather than pure model performance. He pivots to...
Datadog Digs Down Into GPU Efficiency as AI Costs Soar
Datadog has integrated GPU monitoring into its observability platform as AI workloads drive up cloud compute costs. The vendor reports GPUs now account for 14% of cloud spend, and IDC forecasts AI infrastructure spending reached $89.9 billion in Q4 2025, up 62%...
ML Model Builders Must Expand Beyond Pure Modeling
If your job is "I make the ML model", your job is cooked. You need to become more than that. This is how:
Compute and Data Ingestion Power Modern Deep Learning Breakthroughs
got into the deep learning game at the multiple GPUs stage (well, 2 😂) and now we’re at data center compute sizing. the fundamentals are the same, compute (and therefore data ingestion) is what is causing most of the current...
Clinch Brings AI-Driven Creative Personalisation to Pause Ads Through OpenGlass Partnership
Clinch and OpenGlass have certified Clinch’s ad‑serving tags for pause‑ad formats across major CTV publishers such as Paramount, Tubi, and Plex, enabling AI‑driven creative personalization in real time. The integration collapses the historically fragmented, publisher‑by‑publisher workflow into a single tag...
Images 2.0 Crossed an Unseen Qualitative Threshold
Images 2.0 really got over some important qualitative threshold for me that I didn't know existed.
AI Agent Handles Tasks Autonomously, No Babysitting Needed
This is the real deal. It’s an agent you can actually give work to and then walk away. No babysitting. No constant prompting. Just “here’s the mess—figure it out.” Delegate from @abhshkdz & @yutori_ai feels like the first agent that really understands...
AI Predicts 360° Video Attention via Sound and Vision
AI model predicts human attention in 360-degree videos using both sound and vision https://t.co/wCYUUPp6ts https://t.co/tOmHyhka2W
AI Evolves in Months; Security Lags for Years
AI is advancing in months, while securing infrastructure takes years. This is a growing gap. Industry-wide collaboration shows no single company can manage this risk alone. https://t.co/j0voYvfAxQ
AI Funding Soars, Yet Job Impact Remains Unclear
"AI investment is skyrocketing while AI’s impact on jobs and public perception remains mixed" https://t.co/oZvHZHAlDS #AI
Teaching Models Confidence Reduces Overconfidence, Preserves Accuracy
How do top reasoning models become overconfident? MIT found that RL rewards correct answers w/o considering how sure the model is. By training them to estimate their confidence about each answer, the team boosted uncertainty estimates w/o hurting accuracy: https://t.co/zG3GskBc13
AI Clones Top Recruiters, Delivering Hires 10x Faster
Asendia AI builds AI recruiters for staffing agencies and enterprise companies. It clones their best recruiters into AI agents that match, screen, and submit candidates 10x faster. Congrats on the launch, @LajmiRihab and @zormati_ba! https://t.co/2OHkrmBGLx https://t.co/I8j0pjHDKm
Winning with AI Requires Critical Thinking, Not More Agents
AI Agents are surging. So is Critical Thinking. Not a coincidence. Organizations winning at AI aren't the ones with the most agents. They're the ones who know exactly what to ask them. Critical thinking is becoming the rarest skill on the floor.
Google Cloud & DeepMind Bet Big on Voice Interactions
IMO: It's very interesting to see the approach that @GoogleCloud and the DeepMind team is taking with voice interactions. Very different from the classic #CCaaS approach. The DeepMind team is "very bullish" on voice interactions. #CIO #CMO #CX #Cloud #AI #GoogleCloudNext
Exploring Customer Acceleration with DeepMind on Google Cloud
Next up... a deep dive discussion about customer acceleration and DeepMind. #CIO #AI #Cloud #DeepMind #GoogleCloudNext @googlecloud https://t.co/eJ00fv7RSm
Live AI Workflow Demo: Repurposing Content with Claude
Doing an AI “show and tell” of one of my @claudeai content repurposing workflows on @superpathco at 11 CDT, come hang 👀
Tesla AI Strategy and Q1 Highlights on BBC
Great to be on @BBCWorld with @BenThompsonTV discussing Tesla’s AI Path and the 1Q takeaways 🎯🍿🐂📺 @BBCBreaking @BBCNews https://t.co/mdJQBGacLU