Gemini skills are rolling out, but they don’t sync everywhere yet
Google started rolling out skills to Workspace, which lets teams save and reuse instructions. For now, though, skills don’t sync between the Gemini app and Workspace apps. If your team uses both, check which version of your instructions each one is working from.
ChatGPT changes how it reads company information
OpenAI retired admin-managed sync connectors and is moving Enterprise teams to real-time Google Drive, SharePoint, and Teams apps. Existing permissions still apply. If your team relied on synced company knowledge, ask it a few of your usual questions to make sure it still finds the right sources.
Microsoft says your own agents need watching too
Microsoft’s Project Perception is built to monitor agents working inside companies, with their own identities, permissions, and controls. Its point is that an internal agent can cause problems without ever being hacked. So next time you give an agent more access, be clear about what it can change and who can stop it.
DeepSeek ships open-source Flash model with faster outputs and free search API
DeepSeek just dropped V4.1-Flash-1017, and it’s a straight upgrade: faster, cheaper, and shorter outputs than the previous version. No tradeoffs.
The big idea? It uses a technique called reinforcement learning with self-feedback, meaning the model practices tasks, grades its own results, and improves. Think of it like a student who keeps retaking a test until they nail it. This makes it genuinely good at building full projects from scratch, not just answering questions.
Here’s what’s new and usable right now:
- Model is open-source on Hugging Face and ModelScope
- Search and Fetch APIs are now public at $0.75 per 1,000 calls
- Works with Claude Code, Codex, and DeepSeek Harness (upgrade to v0.2.1-rc.1)
- Search and Fetch calls via server-side tools are free for API users until Oct 31
The Search and Fetch APIs let your app pull live web data without building your own scraper. That’s a big unlock for anyone building research tools or AI agents.
Anthropic’s free scanner found 77 critical bugs in open-source projects in one week
Anthropic just launched OSS Scanner, a free tool that automatically hunts for security holes in open-source projects using its strongest AI models. Think of it as a tireless security intern that never sleeps and never misses a bug.
Here’s the problem it solves: almost all software relies on open-source code, much of it maintained by small teams of volunteers. Those teams rarely have time for deep security audits. OSS Scanner fills that gap.
What you actually get per report:
- A working proof-of-concept that reproduces the bug
- A plain explanation of why it’s dangerous
- A suggested patch where possible
Early results: 77 critical and 803 high-severity findings already reported to maintainers. These included multiple vulnerabilities chained to unauthenticated remote code execution exploits.
To enroll, open a pull request adding a directory under projects/name with a project.yaml file containing your repo address and contact email. Anthropic verifies you’re a core maintainer before accepting.
Open-source tool generates game audio locally in seconds, no API key needed
Audio Studio MCP is a free, open-source tool that generates background music and sound effects directly on your Mac. No cloud, no API key, no cost per generation.
The problem it solves: game audio and video sound design used to mean paid subscriptions or hunting royalty-free libraries. Now you just describe what you want in plain English.
Under the hood it runs Stable Audio 3 (Stability AI’s audio model) locally on Apple Silicon. On an M1 Mac, a 60-second track generates in roughly 5 seconds after the model loads.
Here’s what you can generate and export:
- BGM, instrument loops, and sound effects from one sentence, up to 4 candidates at once
- Edit existing clips: restyle, regenerate a section, or extend a track
- Export as OGG, WAV, or FLAC for Godot or Unity, with loudness normalization baked in
To get started: paste the GitHub repo link into Claude Code and ask it to install and register the MCP server. Or manually run git clone then ./install.sh. Requires Apple Silicon with 16 GB RAM recommended. MIT licensed.
Musk Pledges SpaceX Shares to U.S. Kids
Elon Musk pledged to give one SpaceX share to every child in America, calling it the first of many such gifts. The pledge could involve roughly 72 million shares worth about $10.5 billion at recent prices. Musk has signaled support for the Trump Accounts program but has not detailed how or when the shares would be distributed.
Anthropic Eyes $45B Nscale Compute Deal
Anthropic is negotiating to buy roughly $45 billion of computing capacity from Nscale, the Nvidia-backed data center developer, according to Bloomberg. The deal could close within days and would give Nscale a badly needed anchor customer, shoring up its U.S. IPO ambitions while extending Anthropic’s aggressive multi-vendor compute buildout.
Microsoft Pulls Out of Nscale Project
Microsoft, Nscale’s biggest customer, has terminated a critical U.K. data center deal, according to The Telegraph. The pullout deals a heavy blow to the British neocloud operator’s plans for a U.S. listing and highlights how concentrated customer exposure makes AI infrastructure startups fragile when hyperscalers rethink grid-constrained projects.
Tenstorrent Raises $1.5B to Challenge Nvidia
Jim Keller’s Tenstorrent closed a $1.5 billion Series E to expand its open compute ecosystem, pitching its RISC-V-based AI hardware and open software stack as a real alternative to Nvidia GPUs. The raise is one of the largest yet for an AI chip challenger and signals investors still want a second source.
Nvidia, Dell Back doubleAI’s $200M Round
AI model developer doubleAI raised a $200 million Series A co-led by Lightspeed and Bessemer, with Nvidia and Dell Technologies joining the round. The financing reportedly values the company at more than $1 billion, another sign that strategic hardware players keep seeding model startups that will consume their chips and servers.
Amazon Funds New Indiana Nuclear Plant
Amazon signed an agreement with X-energy and American Electric Power to fund a new nuclear facility in Sullivan County, Indiana, on the site of the defunct Breed coal plant. The project targets up to 960 megawatts for the grid and Amazon’s data centers, deepening Big Tech’s bet on nuclear for AI power.
Amazon Takes Aim at Nvidia’s PyTorch Grip
Amazon launched TorchNeuron, a developer tool designed to make PyTorch apps run smoothly on alternative accelerators such as its own Trainium chips. It is AWS’s most direct attempt yet to loosen Nvidia’s software lock-in, since PyTorch underpins most AI applications and CUDA compatibility has been the moat.
Town Hits Unicorn Status in Four Months
AI assistant startup Town reached a unicorn valuation just four months after launch, announcing a $108 million Series B co-led by Forerunner Ventures and Menlo Ventures. The speed of the round shows how hard Bay Area investors are still chasing consumer-facing AI agents that gain early traction.
Eric Schmidt Reportedly Owns 1% of Anthropic
Former Google CEO Eric Schmidt owns roughly 1% of Anthropic, according to Bloomberg, following an early investment in 2021. His stake could be worth around $20 billion if the company’s anticipated IPO reaches a $2 trillion valuation, highlighting the enormous potential windfalls for early AI investors.
Europe Takes On Starlink
Deutsche Telekom, Orange, Telefónica and Vodafone signed a letter of intent to consider a joint bid for spectrum supporting a European direct-to-device satellite service. The carriers would handle ground infrastructure, positioning the alliance as a homegrown rival to SpaceX’s Starlink as Europe pushes for space sovereignty.
Disney Pushes Character.AI Toward Bankruptcy
Disney asked Character.AI to file for bankruptcy as a way to escape mounting lawsuits, according to Wired, while seeking its trust and safety technology and other IP. Dozens of states allege the startup’s chatbots encouraged minors toward self-harm, making the case a test of how AI liability gets resolved.
Agility Unveils Redesigned Digit 5 Humanoid
Agility Robotics detailed its redesigned Digit 5 humanoid, built around safety and practical warehouse work. The company says the robot is designed to operate alongside human workers, handle demanding industrial tasks and support up to 20 hours of daily operation, prioritizing reliable deployment over flashy demonstrations.
Quantinuum Cuts Logical Error Rate 10x
Quantinuum reported a world-record quantum error-correction result, reaching 99.9995% logical fidelity on its Helios commercial computer, a tenfold reduction in logical error rate. The milestone strengthens the Nasdaq-listed company’s case that fault-tolerant quantum computing is arriving faster than skeptics expected.
JPMorgan Deploys $5B Into Defense Tech
JPMorgan Chase has already deployed $5 billion off its balance sheet into direct defense investments, from robot factories to hypersonic missiles and critical supply chains. The bank’s push puts Wall Street money alongside Silicon Valley venture firms chasing the defense tech boom led by startups like Anduril.
Sanofi Taps ScionxBio for Long-Acting RNA
ScionxBio, freshly spun out of Eleven Therapeutics, signed its first partnership with Sanofi to screen the drugmaker’s mRNA candidates using its long-acting RNA platform. The deal shows big pharma still hunting for ways to make mRNA medicines last longer, a key hurdle beyond vaccines.