The Austin-based company will invest in agentic runtime security capabilities to secure AI coding agents.
The post HiddenLayer Raises $100 Million for AI Runtime Security appeared first on SecurityWeek.
https://www.securityweek.com/hiddenlayer-raises-100-million-for-ai-runtime-security/
The startup’s firewall evaluates AI skills, plugins and MCP servers for malicious instructions, excessive permissions and software supply chain risks.
The post AI Agent Firewall Startup AIR Security Emerges From Stealth With $50 Million appeared first on SecurityWeek.
https://www.securityweek.com/ai-agent-firewall-startup-air-security-emerges-from-stealth-with-50-million/
Nvidia has agreed to buy Hugging Face for $12.9 billion, the chipmaker’s second-largest acquisition ever and a direct push into the open-source AI ecosystem it has spent years supplying with hardware.
The deal, disclosed in an SEC filing and a blog post from CEO Jensen Huang, includes roughly $11.9 billion payable to Hugging face shareholders and up to $1 billion in retention equity for employees joining Nvidia. It’s expected to close in the first half of 2027, pending regulatory approval.
Hugging Face has become the default hosting ground for open-source AI: More than 18 million developers use the platform to share upwards of 2 million models, 500,000 datasets, and 1 million applications, Huang wrote in the blog post published today.
Huang framed the acquisition as a commitment to keeping Hugging Face neutral rather than folding it into Nvidia’s stack. Hugging Face will “remain an open platform for the entire AI ecosystem,” he wrote, adding that developers will still choose their own models, frameworks, cloud, and inference providers.
“Together, we will scale Hugging Face’s platform, strengthen its infrastructure, and expand access to AI for developers and institutions worldwide,” Huang wrote.
The acquisition also lands as Nvidia’s biggest customers work to reduce their reliance on GPUs. Anthropic and OpenAI are both developing proprietary chips, and Nvidia has been striking deals across the stack to keep its footprint wide—incluidng a $20 billion listening agreement with chip startup Groq late last year.
Nvidia has tried to buy in before. The company offered $500 million for a stake in Hugging Face last year, valuing it at $7 billion—an offer Hugging Face turned down, according to the Financial Times.
This time, Hugging Face said yes to a full buyout instead. Hugging Face’s annualized revenue is around $150 million, according to The Information, making the price tag a steep multiple on sales—a bet by Nvidia that owning that platform is worth more than the revenue it currently produces.
The timing also follows a security breach last month that put Hugging Face’s infrastructure in the spotlight, after OpenAI’s models went rogue and hacked the open-source platform.
ADWEEK has reached out to Nvidia and Hugging Face for comment.
https://www.adweek.com/dealroom/nvidia-confirms-purchase-of-hugging-face-for-nearly-13-billion/
The CTV campaign used two agents, one for planning and one for buying https://www.adweek.com/media/wellness-brand-rouge-care-let-ai-agents-buy-its-tv-ads-and-made-5x-its-money-back/
The security tool intercepts potentially dangerous agent actions, blocking clear threats and requesting human approval when intent is uncertain.
The post OpenLeash Adds a Human Check to Risky AI Agent Actions appeared first on SecurityWeek.
https://www.securityweek.com/openleash-adds-a-human-check-to-risky-ai-agent-actions/
Google hasn’t released a frontier-level Gemini Pro AI model since early 2026, but it sure loves rolling out new Gemini Flash variants. Today, Google is announcing its third Flash model release in just six weeks, making it more likely that we’ll never see the promised Gemini 3.5 Pro. But no matter, says Google, because Gemini 3.8 Flash is its best reasoning and coding model yet.
Gemini 3.8 Flash comes in two variations. There’s the standard Flash, which Google describes as a “workhorse” model that’s good for anything from agentic tasks to software development. Then we have Gemini 3.8 Flash Cyber, which runs on the same foundations but has been tuned for vulnerability detection and mitigation.
For developers, Google has the same pitch as it did for the 3.7 Flash release just a couple of weeks ago. API access to the model is available at an “introductory rate” through the end of the year: $0.75 per million input tokens and $3.75 per million output tokens. The regular price will be $1.50 / $7.50, but it’s likely there will be new models available long before the price changes. Google probably sees the lower prices as a necessity given that other AI labs have recently dropped token pricing to keep increasingly wary businesses engaged with AI tools.
https://arstechnica.com/ai/2026/09/google-releases-gemini-3-8-flash-its-third-flash-model-in-six-weeks/
Anthropic introduced Enterprise Frontier Safeguards (EFS), a system that combines zero data retention with automated monitoring for misuse.
The post Anthropic Details Response to Security Incidents, Unveils Enterprise Safeguards appeared first on SecurityWeek.
https://www.securityweek.com/anthropic-details-response-to-security-incidents-unveils-enterprise-safeguards/
The designation applies when a model can independently find and exploit zero-day vulnerabilities across many well-defended systems.
The post OpenAI’s Astra Crosses ‘Critical’ Cyber Threshold After Finding Zero-Days appeared first on SecurityWeek.
https://www.securityweek.com/openais-astra-becomes-first-model-to-cross-critical-cybersecurity-threshold/
Brands, agencies and AI labs confront the less glamorous parts of the buildout—performance, control and who’s actually accountable when it breaks. https://www.adweek.com/media/adweek-tech-advantage-ais-honeymoon-with-advertisers-gets-a-reality-check/
While much of the visual media generated by AI has flown under the legal radar, the music industry is famously litigious. Suno is being sued by Universal and Sony for copyright infringement in the US, and a German court recently found the company was violating the country’s music licensing laws. Both cases could lead to hefty fines.
Suno is also in hot water after a hack in late 2025 that showed it scraped content from YouTube, Deezer, and others to train its models. Suno did not disclose the hack, which may have included the private information of millions of users. It’s being sued in Massachusetts over that.
None of that is a good look, so it’s no surprise Suno is looking for ways to change the conversation. The company is essentially positioning itself as a tool for musicians and personal projects, not as a way to generate unlimited slop for Spotify. Shulman stresses in the blog that Suno supports the artistic process, noting that AI technology “can’t replace the human experiences, emotions, and imperfections that make music meaningful.”
That, he claims, is why Suno already has rules designed to guard against misuse. For example, Suno doesn’t allow users to include specific artists or songs in their prompts, and it partners with companies like Musixmatch to ensure people don’t upload copyrighted content to use as a base for their generative compositions. Suno has also strengthened prohibitions against this kind of misuse in an update to its usage policy.
More changes are coming, too. Last year, Suno agreed to limit downloads in a settlement with Warner Music Group. Suno is still working out the specifics, but Shulman says most users won’t be affected. However, limits on downloads, along with watermarking tracks, will apparently help to limit “large-scale abuse.”
Whether these changes will save Suno from its legal woes remains to be seen.
https://arstechnica.com/ai/2026/08/suno-hopes-to-go-legit-with-watermarks-for-ai-generated-music/
![St. Petersburg, FL — December 16th, 2025 — Next Net has launched its breakthrough AI discoverability platform that helps brands and publishers stay visible, cited, and monetizable in AI-driven search. […]](https://rassegna.lbit-solution.it/wp-content/uploads/2026/09/nvidia-confirms-purchase-of-hugging-face-for-nearly-13-billion.png)
