Nvidia OpenAI and Google lead AI news

Aries has raised $7 million in seed funding to build an AI-powered stock trading canvas. The platform will offer real-time market data and personalized recommendations to help traders make better decisions. Aries plans to use the funding to develop new features and grow its user base ahead of its upcoming launch.

Nirva, a startup building wearable AI personal companions, secured $25 million in funding from investors in the US and Europe. The company will use the money to develop its technology and expand its team before its November launch. The AI companions will provide users with personalized recommendations and insights.

Bill Ackman and Eric Schmidt backed Israeli AI startup Two Delta, which raised tens of millions in a seed round. Founded in 2025, Two Delta develops technology to speed up AI models and improve computing efficiency. Its DCOCO technique achieved 2.03 times faster throughput than standard decoding on Nvidia hardware. The company also published EvoWeb, a benchmark showing that AI coding agents struggle to maintain web applications over multiple changes.

OpenAI plans to bring its latest large language models to Google Cloud's Model Garden service. The first model is GPT-6 Astra, which scored perfectly on ExploitBench and includes safety optimizations for enterprise use. Two smaller versions, GPT-6.1 Sol and GPT-6 Luna, cost five and 100 times less than Astra respectively. The models will become generally available within weeks and will later support Google's Confidential Compute, Antigravity, and Gemini Agent products.

Meta shares dropped 6% after quarterly results, contributing to a broader decline in semiconductor and AI trade stocks on October 22. Wafer equipment maker Lam Research also fell during the session. Tesla shares recovered early trading losses. The market pressure followed recent quarterly earnings reports from major tech companies including Nvidia.

A new report highlights the growing importance of pre-execution security for AI-generated code. Security teams must check unfamiliar code before it runs, as generative AI makes it easier for attackers to write malicious code and developers to write new scripts. The report recommends setting clear policies for blocking or restricting code based on risk scores. Earlier checks can reduce the burden on analysts who investigate security alerts.

The SANS Institute announced its 2026 Gulf Region event in Dubai from October 31 to November 26. The program includes 13 specialized courses and two NetWars tournaments covering AI security, threat detection, and industrial control systems. A new course called SEC545 will focus on securing generative AI applications and large language models. The event responds to growing cybersecurity investment in the Middle East, with the regional market expected to reach USD 23.54 billion in 2026.

Rudolf Rojas, a USDA IT official, said agencies cannot always keep a human in the loop for AI tasks due to staffing and resource cuts. He argued that AI agents should operate with minimal privileges under strong risk management and zero-trust network security. Rojas called for new regulations to ensure safe AI use and warned about threats from nation-state actors like China and Russia.

Hong Kong Financial Secretary Paul Chan discussed how the city is attracting top AI talent and firms. He explained Hong Kong's strategy for positioning itself as a leading AI investment and business destination. Chan also addressed the U.S.-China tech rivalry and how Hong Kong offers diversification and safety for investors.

A new open benchmark called OmniParseBench tests document parsers with 16,288 pass-or-fail tests across 2,937 pages. Researchers from Datalab rebuilt the leaderboard from 439,776 verdicts for 27 parsers and verified results by running Tesseract and Indra OCR on real PDFs. The benchmark covers text, reading order, tables, and 92 languages across scanned forms, handwriting, and math documents.

Key Takeaways

  • Aries raised $7M seed funding for an AI-powered stock trading platform offering real-time market data and personalized recommendations.
  • Nirva raised $25M for wearable AI personal companions, with a November launch and investors from the US and Europe.
  • Two Delta, backed by Bill Ackman and Eric Schmidt, achieved 2.03x faster throughput on Nvidia hardware with its DCOCO technique.
  • OpenAI will bring GPT-6 Astra, GPT-6.1 Sol, and GPT-6 Luna to Google Cloud's Model Garden, with Astra scoring perfectly on ExploitBench.
  • Meta shares fell 6% on October 22 amid AI trade pressure, while Tesla recovered early losses and Lam Research also declined.
  • A new report urges security teams to check unfamiliar AI-generated code before execution, recommending risk-based policies and earlier analyst checks.
  • SANS 2026 Gulf Region event in Dubai will feature 13 courses and two NetWars tournaments, including a new SEC545 course on securing generative AI.
  • USDA IT official Rudolf Rojas says keeping humans in the loop for AI is untenable, calling for zero-trust security and new AI regulations.
  • Hong Kong Financial Secretary Paul Chan positions the city as a top AI investment hub offering diversification amid U.S.-China tech rivalry.
  • OmniParseBench tests document parsers with 16,288 checks across 2,937 pages, covering 92 languages and multiple document types.

Aries raises $7M seed for AI stock trading platform

Aries, a stock trading platform, raised $7 million in seed funding to build an AI-powered trading canvas. The tool will offer real-time market data and personalized recommendations to help traders make better decisions. The company plans to use the money to develop new features and grow its user base. The platform is expected to launch in the coming months.

Nirva raises $25M for AI wearable companion

Nirva, a startup building wearable AI personal companions, raised $25 million in funding. The money will help the company develop its technology and expand its team before its November launch. The AI companions will give users personalized recommendations and insights. Investors from the US and Europe took part in the funding round.

AI code makes pre-execution security more important

A new report says security teams must check unfamiliar code before it runs, especially with AI-generated code on the rise. Generative AI can help attackers write malicious code and developers write new scripts, making it harder to spot threats. The report recommends setting clear policies for blocking or restricting code based on risk scores. Earlier checks can reduce the burden on analysts who investigate security alerts.

SANS Gulf Region 2026 focuses on AI cybersecurity training

SANS Institute announced its 2026 Gulf Region event in Dubai from October 31 to November 26. The program includes 13 specialized courses and two NetWars tournaments covering AI security, threat detection, and industrial control systems. A new course called SEC545 will focus on securing generative AI applications and large language models. The event responds to growing cybersecurity investment in the Middle East, with the regional market expected to reach USD 23.54 billion in 2026.

Meta and chip stocks fall as AI trade faces pressure

Stocks declined on October 22 as semiconductor companies fell amid pressure on the AI trade. Meta shares dropped 6% after quarterly results, while wafer equipment maker Lam Research also declined. Tesla shares recovered early trading losses during the session. The market pressure followed recent quarterly earnings reports from major tech companies.

OmniParseBench Guide Tests Document Parsers with Thousands of Checks

A new tutorial explains OmniParseBench, an open benchmark from Datalab for testing document parsers. It uses 16,288 pass-or-fail tests across 2,937 pages to grade parser accuracy. Researchers rebuilt the leaderboard from 439,776 verdicts for 27 parsers and verified results by running Tesseract and Indra OCR on real PDFs. The benchmark covers text, reading order, tables, and 92 languages across scanned forms, handwriting, and math documents.

Bill Ackman and Eric Schmidt Back Secretive Israeli AI Startup Two Delta

Israeli AI startup Two Delta raised tens of millions in a seed round from investors including Bill Ackman and Eric Schmidt. Founded in 2025, the company develops technology to speed up AI models and improve computing efficiency. Its DCOCO technique achieved 2.03 times faster throughput than standard decoding on Nvidia hardware. Two Delta also published EvoWeb, a benchmark showing that AI coding agents struggle to maintain web applications over multiple changes.

USDA IT Official Says Keeping Humans in AI Loop Is Untenable

Rudolf Rojas, a USDA IT official, said agencies cannot always keep a human in the loop for AI tasks due to staffing and resource cuts. He argued that AI agents should operate with minimal privileges under strong risk management and zero-trust network security. Rojas called for new regulations to ensure safe AI use and warned about threats from nation-state actors like China and Russia. He also urged vendors to improve how they implement AI in government tools.

OpenAI Brings Advanced AI Models to Google Cloud

OpenAI plans to make its latest large language models available through Google Cloud's Model Garden service. The first model is GPT-6 Astra, which scored perfectly on ExploitBench and includes safety optimizations for enterprise use. Two smaller versions, GPT-6.1 Sol and GPT-6 Luna, cost five and 100 times less than Astra respectively. The models will become generally available within weeks and will later support Google's Confidential Compute, Antigravity, and Gemini Agent products.

Hong Kong Financial Secretary Pushes City as Top AI Investment Hub

Hong Kong Financial Secretary Paul Chan explained how the city is attracting top AI talent and firms. He discussed Hong Kong's strategy for positioning itself as a leading AI investment and business destination. Chan also addressed the U.S.-China tech rivalry and how Hong Kong offers diversification and safety for investors.

Sources

NOTE:

This news brief was generated using AI technology (including, but not limited to, Google Gemini API, Llama, Grok, and Mistral) from aggregated news articles, with minimal to no human editing/review. It is provided for informational purposes only and may contain inaccuracies or biases. This is not financial, investment, or professional advice. If you have any questions or concerns, please verify all information with the linked original articles in the Sources section below.

AI-powered trading AI companions AI funding AI speedup AI security AI regulations AI talent AI benchmark AI investment AI diversification AI language models AI stock trading AI recommendations AI market data AI zero-trust AI risk management AI document parsing AI open benchmark AI generative code

Comments

Loading...