Salesforce, Microsoft Copilot AI Security, OpenAI Models

Several companies are launching AI security solutions to address emerging risks. SPLX has added runtime protection to its platform, guarding against prompt injections and data leaks in AI chatbots. Noma Security debuted its AI Agent Security solution, designed to secure AI agents across platforms like Salesforce AgentForce and Microsoft Copilot Studio. Tenable unveiled Tenable AI Exposure, a tool for managing risks associated with AI tools like ChatGPT Enterprise and Microsoft Copilot, set for release in 2025. These tools aim to help businesses manage AI usage, prevent data leaks, and control how AI is used. In other news, OpenAI released open-source AI models to compete with China, offering alternatives that allow users to run AI on their own devices. These models, including gpt-oss-120b and gpt-oss-20b, are designed for reasoning and web access. Huawei upgraded its Xinghe AI Fabric for Latin American enterprises, providing a three-layer network design to enhance data center operations. AI is also being used for both beneficial and malicious purposes. The Tolleson Union High School District in Arizona is using AI security cameras to prevent weapons from entering schools, having stopped eight weapons during the 2024-2025 school year. However, AI is also being used for information warfare, with reports of China using AI to monitor public opinion and run influence campaigns. A network of AI-powered bots on X, formerly Twitter, was found to be promoting Donald Trump, though they later split over the Epstein files. In business applications, Salesforce acquired Bluebirds to enhance AI sales prospecting, integrating the technology into its Sales Cloud and Agentforce products. A Microsoft study suggests AI is unlikely to replace real estate agents, but agents should still use AI to their advantage. Meanwhile, Junzhou Huang at The University of Texas at Arlington received a $3.1 million grant to use AI in drug discovery. Kevin Clark was honored for his AI work at Tranquility AI, where he develops AI tools for law enforcement. Insurance compliance leaders are also embracing AI, with companies like The Mutual Group creating guidelines to ensure AI is used fairly and transparently.

Key Takeaways

• SPLX launched a new runtime protection tool to secure AI chatbots from attacks like prompt injections and data leaks.
• Noma Security released its AI Agent Security solution, securing AI agents in platforms like ServiceNow, Salesforce AgentForce, and Microsoft Copilot Studio.
• Tenable's AI Exposure tool, launching in 2025, will help businesses manage risks from AI tools like ChatGPT Enterprise and Microsoft Copilot.
• OpenAI released open-source AI models, gpt-oss-120b and gpt-oss-20b, to compete with China and offer more control to users.
• Huawei upgraded its Xinghe AI Fabric for Latin American enterprises, enhancing data center network reliability and computing power.
• AI security cameras in the Tolleson Union High School District stopped eight weapons from entering schools during the 2024-2025 school year.
• China is using AI for information warfare, including monitoring public opinion and running influence campaigns.
• Salesforce acquired Bluebirds to enhance AI sales prospecting within its Sales Cloud and Agentforce products.
• A Microsoft study indicates AI is unlikely to replace real estate agents due to the importance of emotional connection in transactions.
• Junzhou Huang at The University of Texas at Arlington received a $3.1 million grant to use AI in drug discovery.

SPLX adds AI runtime protection to security platform

Startup SPLX expanded its AI security platform with a new runtime protection tool. This tool helps protect AI chatbots and agents from attacks like prompt injections and data leaks. SPLX also launched Analyze, a feature that summarizes results from its automated red teaming tool, Probe. These tools help businesses secure their AI systems and address concerns about AI risks like data leakage and unreliability. MSSPs and MSPs can also use the SPLX platform to protect their customers' AI environments.

Noma Security launches AI agent security solution for enterprises

Noma Security has released its AI Agent Security solution, which is the first platform that secures agentic AI. The platform helps companies find, watch, and manage AI agents, allowing them to use AI safely. AI agents can have risks like operating without oversight or accessing too much data. Noma's platform protects against these risks and works with existing security systems. It covers different types of AI agents, like those in ServiceNow, Salesforce AgentForce, Microsoft Copilot Studio, LangChain, CrewAI and Github Copilot.

Tenable's new AI security tool manages ChatGPT risks

Tenable has launched Tenable AI Exposure, a tool that helps businesses manage risks from AI tools like ChatGPT Enterprise and Microsoft Copilot. The tool lets companies see how employees use AI, what data is exposed, and how attackers might manipulate the systems. Tenable AI Exposure finds AI usage, manages risks like data leaks, and controls how AI is used. It integrates with the Tenable One platform to provide a complete view of AI risks. The tool is currently in a private preview and will be available later in 2025.

Noma Security debuts AI agent security solution for businesses

Noma Security has launched its AI Agent Security solution, which is the first platform that secures agentic AI. The platform helps companies find, watch, and manage AI agents, allowing them to use AI safely. AI agents can have risks like operating without oversight or accessing too much data. Noma's platform protects against these risks and works with existing security systems. It covers different types of AI agents, like those in ServiceNow, Salesforce AgentForce, Microsoft Copilot Studio, LangChain, CrewAI and Github Copilot.

AI bot network promoting Trump split over Epstein files

A network of AI-powered bots on X, formerly Twitter, that supported Donald Trump became divided over the release of the Epstein files. Alethea, an analytics firm, discovered at least 400 fake X profiles using AI to generate content. These bots initially praised Trump but later disagreed about his handling of classified files related to Jeffrey Epstein. The bots used 'PromptPasta,' creating varied responses to user inputs, making them harder to detect. Experts say platforms need to enforce policies against spam and scams, as AI lowers the barrier to creating sophisticated influence campaigns.

China uses AI for information warfare, monitor public opinion

The Chinese government is using AI to monitor and influence public opinion, giving them a new tool for information warfare. A company called GoLaxy uses AI to run influence campaigns and collect data on politicians. American spy agencies are watching GoLaxy for signs it might try to influence American elections. The company's technology can track public debates and create propaganda to change public opinion. GoLaxy denies these claims, but documents show it can target individual social media users with customized content.

OpenAI releases open-source models to compete with China

OpenAI has released two open-source AI models to compete with China in the AI market. These models allow customers to run AI directly on their devices, saving costs and protecting privacy. OpenAI hopes these models will appeal to countries wanting more control and independence from cloud providers. The models are designed for reasoning and web access. Industry leaders say it's important for the U.S. to stay competitive in open-source AI.

OpenAI releases first open AI model in years

OpenAI has launched its first open AI model since 2020, offering two versions for developers. These models aim to advance reasoning in open-source AI. OpenAI's CEO admitted the company was wrong to limit access to its models. The new models, gpt-oss-120b and gpt-oss-20b, target AI developers for real-world applications. The larger model can run on a single Nvidia 80GB chip, while the smaller one fits on a Mac laptop. OpenAI did not compare its models against Chinese open-source systems.

AI security cameras stopped weapons at Valley school district

The Tolleson Union High School District in Arizona is using AI security cameras to keep weapons off school campuses. During the 2024-2025 school year, the cameras stopped eight weapons from entering schools. The district has about 150 to 200 cameras watching 14,000 students. The cameras help staff react to incidents like fights. The district is considering adding AI weapons detectors to work with the existing cameras.

Kevin Clark honored for AI excellence at Tranquility AI

Kevin Clark is recognized in Marquis Who's Who for his work in AI. He has 25 years of experience in analytics and digital forensics. As VP at Tranquility AI, he develops AI tools for law enforcement. Previously, he led a crime strategies unit for Colorado's 4th Judicial District Attorney's Office. He also served in the Air Force for nearly a decade. Clark is involved in community initiatives and lectures at Pikes Peak State College.

Insurance compliance leaders embrace AI innovation

As AI changes the insurance industry, compliance leaders are taking on new roles. Shannon Woods, chief compliance officer at The Mutual Group, says it's important to create strong guidelines that allow innovation while maintaining trust. Woods believes companies should not wait for perfect regulations before using AI. The Mutual Group uses AI in areas like claims processing and customer service. They have an AI governance program to ensure AI is used fairly and transparently.

AI research aims to speed up drug development

Junzhou Huang at The University of Texas at Arlington received a $3.1 million grant from the National Institutes of Health to use AI in drug discovery. His research focuses on using machine learning to design antibodies that bind to viruses and other antigens. This could significantly speed up the process of developing treatments for diseases. Huang's team previously ranked high in an international AI challenge for protein structure prediction. He also received an award from Johnson & Johnson to explore AI-based toxicology prediction.

AI won't replace real estate agents, says Microsoft study

A Microsoft study suggests AI is unlikely to replace real estate agents. While AI can handle tasks like customer service and sales, it struggles with the emotional aspects of real estate transactions. Buyers and sellers want a trusted advisor, something AI cannot easily provide. However, real estate agents should still use AI to their advantage. Agents who don't use AI risk losing business to those who do.

Huawei upgrades AI Fabric for Latin American enterprises

Huawei has upgraded its Xinghe AI Fabric solution for Latin American businesses. The solution uses a three-layer network design with AI Brain, AI Connection, and AI Network Element. It aims to provide a reliable data center network with full computing power and continuous service. The upgrade includes features like network automation, optimized path planning, and high-precision traffic sensing. Brazil Compwire is using Huawei's solution to build efficient networks and improve data center operations.

Salesforce buys Bluebirds to enhance AI sales prospecting

Salesforce is acquiring Bluebirds, an AI-driven prospecting system, to improve lead generation. Bluebirds' system helps sales teams find and engage with high-potential leads more effectively. The technology will be added to Salesforce's Sales Cloud and Agentforce products. This acquisition will help small businesses compete with larger companies by automating the prospecting process. Bluebirds also offers analytics to help sales teams adjust their strategies.

Sources

• Startup SPLX Expands AI Security Platform with Runtime Protection Tool
• Noma Security Launches Industry's First Comprehensive AI Agent Security Solution
• New Tenable AI Security Platform Helps Enterprises Control ChatGPT Risks: What Security Teams Need to Know
• Noma Security Debuts Industry's First Comprehensive AI Agent Security Solution
• How the Epstein Files Blew Up a Pro-Trump AI Bot Network on X
• China Turns to A.I. in Information Warfare
• AI execs back OpenAI's open-source return
• Fortune Tech: Indecipherable
• Valley school district said AI security cameras stopped 8 weapons from entering campuses last school year
• Marquis Who's Who Honors Kevin Clark for Excellence in Artificial Intelligence
• Waiting for AI clarity? "You're already behind"
• AI research targets faster drug development
• What jobs will AI replace? Not real estate agents (probably)
• Huawei Releases Newly Upgraded Xinghe AI Fabric Solution, Empowering Latin American Enterprises with Full Computing Power and Always-on Services
• Salesforce acquires Bluebirds to boost AI-driven sales prospecting