AI News Archive: July 2026 - Monthly Highlights
74 articles published this month
Nous Research in Talks for New Funding at USD 1.5 Billion Valuation
Nous Research is closing in on a new funding round that would value the AI startup at $1.5 billion, according to three people familiar with the deal.
ACRouter's AI model selector beats Opus-only setups by 2.6x on cost
Enterprises running AI at scale rarely stick to one model. They split traffic across GPT-5.5 for hard reasoning, cheaper open-weight models like Kimi...
Security Researchers See Attackers Use Prompt Injections to Disable AI Defenses
Prompt injection has been the attacker's favorite trick for more than two years now: bury a command in an email, a calendar invite, or a webpage, and...
Nadella calls AI labs "ironic" for banning distillation while using others' data
Satya Nadella has a problem with how his rivals write their terms of service. In a post on his personal blog, the Microsoft CEO singled out AI labs,...
Google's SensorFM outperforms models on 34 of 35 health data tasks
Google Research trained a foundation model on more than one trillion minutes of wearable sensor data pulled from five million Fitbit and Pixel Watch...
German AI Consortium's Soofi S Hits Benchmarks, Generates 8x Faster
A German research consortium coordinated by the KI Bundesverband has released Soofi S 30B-A3B, an open-source language model trained entirely on...
Researchers Propose Method to Block Illegal AI Content for Kids
The National Center for Missing and Exploited Children logged more than 1.5 million reports of AI-generated child sexual abuse material in 2025, up...
NVIDIA Toolkit Accelerates OpenFold3 Co-Folding Workflow
Virtual screening runs against millions to billions of compounds, and co-folding models like OpenFold3 often produce the most accurate structures in...
OpenAI Launches ChatGPT Work, a Unified AI Agent Powered by GPT-5.6
OpenAI rolled out ChatGPT Work on Thursday, a new agent built into its chatbot that's meant to do actual jobs rather than just answer questions.
OpenAI's GPT-5.6 Sol Ultra Solves 50-Year-Old Math Problem in an Hour
OpenAI says its latest model, GPT-5.6 Sol Ultra, has produced a complete proof of the Cycle Double Cover Conjecture, a graph theory problem that sat...
Terrorist Groups Use Major AI Chatbots for Attack Planning, Weapons Development
Boko Haram has an AI training program, and ISIS built it. That's the core finding in a new study from Antonia Jülich, a researcher with the Cambridge...
OpenAI Targets Families as ChatGPT Adapts for Household Use
OpenAI is hiring a product manager in San Francisco to build features aimed at families, caregivers, and older adults, according to a job posting the...
Meta's Muse Spark 1.1 coding score hits 71.3, edges past GLM-5.2
Meta pushed out an update to its Muse Spark model on Thursday, and the numbers put it ahead of Zhipu's GLM-5.2 on coding tasks for the first time.
Apple Sues OpenAI, Alleges Trade Secret Theft for Hardware
Apple filed suit against OpenAI on Friday in the U.S. District Court for the Northern District of California, accusing the company of stealing trade...
Kyutai Releases MuScriptor AI for Multi-Instrument Music Transcription to MIDI
Automatic music transcription has always run into the same wall: a solo piano recording turns into clean MIDI without much fuss, but hand it a full...
Anthropic's 'Logit Lens' Reveals How Claude Puzzles Over Words
Anthropic says it has found a way to watch Claude think before it speaks, literally, at the level of individual words.
Governments Rush to Use AI for Cyber Defense Despite Risks
An AI agent built to defend a network can be turned into the tool that breaks in.
NVIDIA’s New Nemotron Model Achieves 2.03x Server Throughput
NVIDIA's Nemotron-3-Super has a familiar problem: it's accurate, but it's a hybrid Mamba-Transformer MoE with 120.7B total and 12.8B active...
OpenAI Releases GPT-Live Voice Models That Delegate Reasoning to GPT-5.5
OpenAI shipped a new voice model family Tuesday called GPT-Live, and it now runs the ChatGPT Voice experience for every user, everywhere, starting...
Bezos-backed startup raises USD 320 million to build AGI from gaming data
General Intuition just closed a $320 million funding round at a $2.3 billion valuation, with Jeff Bezos, Eric Schmidt, Coatue, and researchers from...
ZML releases free tool to speed AI inference across mixed chips
ZML, a Paris-based AI startup backed by Turing Award winner Yann LeCun, has released a free inference server called ZML/LLMD built to run open-source...
Meta's Muse AI model now generates images for Instagram, WhatsApp
Meta rolled out its first image generator built entirely inside Superintelligence Labs on Tuesday, and it's already live in the Meta AI app,...
Anthropic Launches Claude Cowork AI Agent for Mobile and Web
Anthropic is taking Claude Cowork off the desktop leash. The AI agent, which launched as a desktop-only feature, is now rolling out to mobile and...
Anthropic Moves Its Claude Agent to Phones as AI Rivals Target Mobile
Anthropic no longer wants Claude Cowork chained to a laptop lid. The company announced Tuesday that its agent, which handles digital busywork like...
DeepSeek Plans Own Chips Amid US Export Controls, Following Huawei, Alibaba
DeepSeek wants to build its own chips. Reuters reported on July 7, citing three people familiar with the matter, that the Chinese AI startup has...
AI chemist proposes plans; humans pick four for lab testing, boosting reaction
Forget the hype about AI designing drugs. The real work is much dirtier, happening one stubborn chemical reaction at a time.
Hanns Christoph Nägerl’s team finds quantum heating defies classical intuition
Motors scream. Pans sizzle. A child on a swing arcs higher with every push. Hanns Christoph Nägerl's lab at the University of Innsbruck just broke...
Trump’s Davos drama, AI‑fuelled midterms draw tens of millions early
The money is already flowing, tens of millions, eleven months out from the midterms. That’s not normal.
Sen. Warren urges Trump admin to expand AI credit after OpenAI letter
Senator Elizabeth Warren is demanding answers from the Trump administration. Her focus: a quiet request from OpenAI.
AI-Run Ransomware Attack Still Required Human Involvement
Sysdig researchers said last week they'd found the first documented case of "agentic ransomware," an extortion campaign called JadePuffer where an AI...
Dynamic Power Boost Compensates for GPU Failures in LLM Training
A training run spread across several thousand GPUs doesn't fail all at once. It fails a little at a time: a node drops, a link flakes, a device goes...
Vercel CEO: Eve AI Agents Can Improve Productivity Across Company
Vercel now handles 6 million deployments a day, and half of them come from coding agents rather than human developers pushing code.
Zhipu AI Launches ZCode With 1M-Token Context Window
Zhipu AI's Beijing-based Z.ai has a new coding tool called ZCode, built on its GLM-5.2 model, and it's aimed squarely at Claude Code and OpenAI's...
Tencent's Hy3 Model Matches Larger Rivals With 21 Billion Parameters
Tencent put out Hy3 this week, its latest open-source language model, and the numbers on paper are meant to make bigger competitors nervous.
16B-Parameter Diffusion Models Generate Multi-Speaker, Multilingual Speech
Apple researchers have trained a continuous diffusion speech model with 16 billion parameters on tens of millions of hours of conversational audio,...
Specialized Model Cuts ASR Errors With 15x Fewer Parameters Than LLMs
A team of six researchers, including Zijin Gu, Tatiana Likhomanenko and Navdeep Jaitly, is pushing back on the assumption that fixing speech-to-text...
PathMoE Study Shows More Concentrated, Robust Routing Paths
A team at Apple, including Zijin Gu, Tatiana Likhomanenko, Vimal Thilak, Jason Ramapuram, and Navdeep Jaitly, has published new work on how sparse...
New Research Shows Why Agent Rankings Change After Accounting for Competition
A team testing agent configurations ran into a familiar problem: change the model, rewrite the prompt, swap a retrieval tool, and the average score...
China Orders ByteDance, Alibaba to Shut Down Custom AI Chatbots
ByteDance's Doubao chatbot has more than 300 million monthly users, making it the most popular AI assistant in China.
Five Small, Open-Weight Models Built for Agentic Tool Calling
Five small, open-weight models built specifically for tool calling landed in agent pipelines this year, and none of them come from the usual frontier...
Amazon Mechanical Turk enters maintenance mode as AWS retires data labeling services
Amazon Web Services will stop accepting new customers for Mechanical Turk starting July 30, 2026, according to an announcement posted through AWS...
Meta Tests GPT-5.5 With 'Watermelon' Model
Meta's Muse Spark model launched in April to a shrug. Industry read: usable, but nowhere near frontier.
Trump cracks down on Anthropic after Amazon tip; staff largely foreign
The Trump administration just banned Anthropic. Amazon provided the tip. The stated reason is the company's largely foreign workforce.
GPT-4o, Gemini, and Claude Vision Can Reason Over Visual Details
Ask GPT-4o to look at a photo of a crowded whiteboard and it won't just tell you there's a whiteboard in the room.
AI Search Agents Struggle With Ambiguous Queries, Study Finds
Artificial intelligence will answer your questions before you finish asking them. It's good at that.
pxpipe hides text in PNGs to cut Claude token costs by up to 70%
Everyone's hunting for cheaper AI bills. Now there's pxpipe. This new open-source tool employs a brutally simple hack: hide your text inside a...
Midjourney Challenges Studios' AI Document Secrecy in Court Filing
Hollywood studios suing AI companies for copyright infringement is a clean, righteous story.
Run Local AI on 8GB Macs With Smaller Models, Avoid Complex Setup
The fantasy of running your own AI isn't about freedom or digital sovereignty. It's about wanting to type a stupid question about lunch without it...
Typed Answer Contract Prevents RAG Hallucination With Programmatic Signals
Hallucination is not a bug in RAG, it is generative AI’s default behavior. Large language models predict the next token; they do not look things up.
Deep Learning AI Models Identify Data Features Without Human Input
Forget the jargon. Modern AI has learned to see. That's the fact. For decades, a human expert had to tell a computer what to look for.
AI Agent Skips Unneeded Tool Call After Observing Zero Precipitation
Most AI assistants are obedient, expensive idiots. They follow the script, run every function you give them, and rack up your API bill.
Long Context Models Reduce Compute Waste by Eliminating Padding
Bigger context windows sell chips. The tech industry has obediently chased that line, from a few hundred tokens to thousands, treating length as an...
Developer Replaces LLM Wiki With Pure Python Compiler, Citing Over-Engineering
We keep turning simple file folders into expensive AI projects. One developer just turned theirs back.
Alibaba Bans Employees From Using Claude AI Amid China Restrictions
Alibaba told its tech workforce to delete Claude this week. A direct order. The American AI assistant is now forbidden, joining a blacklist that...
Meta's AI Agent Push Slower Than Planned After Workforce Restructuring
Mark Zuckerberg told his own employees the plan isn't working. His big AI bet is stalling.
Wiola Architecture Introduces Five Novel Components for Efficient Small Language Models
The obsession with trillion-parameter models has become a bad joke. A useful one is happening with small ones.
Agent4cs Uses Multi-Agent System for Hierarchical Code Summarization
Every developer knows the dread of a massive, unfamiliar codebase. You're trying to reconstruct an architect's mind from a million scattered...
Auto-FL-Research Uses Agents to Automate Federated Learning Algorithm Search
Building a better federated learning algorithm is a grind. Researchers must make a cascade of small, interlocking decisions—how clients update...
t0-alpha Shows Tight 0.015 CRPS Spread in Time-Series LLM Cluster
A new kind of AI model is starting to predict the future, or at least next quarter's sales and next week's server load.
VideoFlexTok's Flow Decoder Enables Variable-Length Video Tokenization
Video models are stuck on a bad idea. They treat every second of footage the same, forcing the same dense grid of data tokens onto a simple scene and...
AI Engineers Face Rising Costs, Need New Strategies for Efficiency
Engineers are now judged by their AI appetite. Teams track token counts, and some have leaderboards.
60% of Experts Say Humanity's Last Exam Is Necessary and Useful
Most AI benchmarks are useless now. The good ones are too easy. The latest large language models score over 90% on tests that were considered hard...
Square's ChatGPT integration charges restaurants 6% fee for pickup orders
Restaurants are about to start getting orders from chatbots. The economics might actually work this time.
Enterprise AI Governance Relies on Manual Monitoring, Survey Finds
Companies are adding AI tools much faster than they are setting up systems to manage them, according to a new survey.
Z.ai launches ZCode to challenge GitHub Copilot, Claude Code
Another AI coding tool just launched. This one is Chinese. It matters because nothing in this market is simple anymore.
New Framework Shifts LLM Output to Typed JSON for Safer Web Data Collection
LLMs can't scrape websites. They can write code that attempts to scrape websites, and that code usually fails.
Gemini Update Adds Screen Reactions, AI Video Creation in June 2026
Google's latest AI update is a kitchen sink of features, promising to react to your screen, make videos, and babysit your portfolio.
Random Split Identified as Most Leakage‑Prone in Spatial‑Temporal Prediction
Why does a random split matter in spatial‑temporal forecasting? Because geography isn’t a tidy list of independent rows.
Anthropic adds security measure; Commerce Dept clears Fable 5 for release
The Trump administration lifted export controls on Anthropic’s Claude Fable 5 after the company agreed to add a new guardrail.
Study Evaluates AI Retrieval Techniques for Finding Models Across Formats
Searching for a specific simulation model today is a brute-force nightmare. It's like hunting for one uniquely stamped brick in a vast, unmarked...
Claude Sonnet 5 lists unchanged token rates despite double real cost
Anthropic rolled out Claude Sonnet 5 this week. Check the pricing page: the per-token rates haven't budged.
Anthropic's Claude Code hides XOR‑encrypted flag for Chinese users in v2.1.91
Anthropic hid a simple trap in its code. Version 2.1.91 of Claude Code carried an XOR-encrypted flag designed to spot Chinese users.
NVIDIA launches Nemotron‑Labs‑TwoTower diffusion model with 128‑expert MoE
NVIDIA just dropped a diffusion language model that thinks in stereo. Nemotron‑Labs‑TwoTower marries a frozen autoregressive backbone with a second...
Anthropic launches Claude Science, expanding flagship tools for coders
Anthropic just turned its code-writing bot into a lab assistant. The company announced Claude Science on Tuesday, a new flagship tool built to parse...