google, microsoft and chatgpt Updates

Recent developments highlight both the promise and perils of AI across various sectors. Curity CTO Jacob Ideskog warns of security risks from AI agents in enterprise systems, drawing parallels to early cloud adoption's security oversights. This concern is validated by a vulnerability in Lenovo's GPT-4 powered chatbot, where a simple 400-character prompt could enable attackers to steal session cookies, emphasizing the need for robust security measures. Google Cloud is responding with new AI security features, including automated AI agent discovery, real-time threat protection via Google Cloud Model Armor, and threat detection using Mandiant intelligence, alongside AI consulting services. They're also introducing an AI-powered security assistant to automate tasks and improve AI system protection. In education, Rogers State University is launching AI and education degrees, while Topeka Public Schools are using AI to assist teachers in lesson planning. Pennsylvania is expanding AI use in government, leveraging ChatGPT Enterprise to save employee time on routine tasks. Nitrogen has launched AI Meeting Center for financial advisors, integrating with platforms like Zoom and Microsoft to automate meeting notes. However, AI's risks are also apparent in mental health, where tools can pose manipulation and psychological harm, necessitating ethical safeguards. Robinhood has launched Digests by Robinhood Cortex, an AI-powered investing tool, in the UK to explain stock movements. Finally, AI's potential for misuse is highlighted by the spread of fake hurricane photos online, underscoring the need for trusted information sources.

Key Takeaways

• Curity CTO Warns Curity that AI agents in enterprise systems pose security risks, similar to early cloud adoption.
• Lenovo's GPT-4 chatbot had a vulnerability allowing attackers to steal session cookies with a 400-character prompt.
• Google Cloud has released new AI security features, including automated AI agent discovery and real-time threat protection with Google Cloud Model Armor.
• Rogers State University is introducing bachelor's degrees in elementary education and artificial intelligence.
• Nitrogen launched AI Meeting Center for financial advisors, integrating with Zoom and Microsoft.
• Topeka Public Schools are using AI to help teachers create lesson plans.
• Pennsylvania is expanding AI use in government agencies, using ChatGPT Enterprise to save employee time.
• AI mental health tools pose risks like manipulation and psychological harm, requiring ethical safeguards.
• Robinhood launched Digests by Robinhood Cortex, an AI-powered investing tool, in the UK.
• Fake hurricane photos generated by AI are spreading online, highlighting the need for trusted information sources.

AI Agents Pose Security Risks Warns Curity CTO

Jacob Ideskog, CTO of Curity, warns that AI agents in enterprise systems create security risks like data leaks and unauthorized access. He compares the current situation to early cloud adoption, where security was often overlooked. Ideskog advises companies to implement safeguards like prompt hardening and continuous monitoring. He cites examples like Cursor IDE and GitHub Copilot where AI tools introduced vulnerabilities, emphasizing the need for a new approach to AI security.

Lenovo Chatbot Hack Exposes AI Security Weakness

A security flaw in Lenovo's GPT-4 powered chatbot allowed attackers to steal session cookies. Cybernews researchers found the chatbot was open to cross-site scripting attacks because it didn't clean up inputs and outputs correctly. Attackers used a 400-character prompt to inject malicious code. Experts warn that companies are quickly using AI chatbots without proper security, treating them as experimental instead of critical applications. They recommend using the same security measures as with web applications and staying updated on prompt engineering best practices.

Lenovo AI Chatbot Vulnerability Allows Remote Code Execution

Researchers found that Lenovo's AI chatbot, Lena, had security problems that could let attackers run malicious code and steal data. The chatbot, which uses OpenAI's GPT-4, was open to cross-site scripting attacks. A simple 400-character prompt could trick the chatbot into creating harmful HTML responses. This could allow attackers to steal session cookies and access Lenovo's systems. Lenovo has fixed the problem, and experts recommend checking all chatbot outputs and using strict security measures.

Google Cloud Adds New AI Security Tools

Google Cloud has released new AI security features to help businesses protect their AI projects and use AI to improve their overall security. These tools include automated discovery of AI agents, real-time protection against threats like prompt injection using Google Cloud Model Armor, and threat detection for AI agents using Mandiant intelligence. Google Cloud also offers AI Consulting services for risk management and security planning. The goal is to make security easier and more effective for companies using AI.

Google Cloud Introduces AI-Powered Security Assistant

Google Cloud is using AI to help security teams by automating tasks and protecting AI systems from attacks. The company is improving its AI Protection solution in Security Command Center to automatically find AI agents and servers, identify security problems, and block threats like prompt injection. Google is also introducing an Alert Investigation agent that uses AI to analyze security events and suggest actions. These tools are designed to free up security experts to focus on important issues.

Rogers State University Adds AI and Education Degrees

Rogers State University (RSU) now offers bachelor's degrees in elementary education and artificial intelligence. The AI degree is an option within the information technology program. RSU is also adding a master's degree option to its cybersecurity and nursing programs. The new AI program will teach students how large language models work and how to build their own AI systems. RSU aims to equip students with skills needed in the growing AI field, while also addressing the ethical considerations of AI.

Nitrogen Launches AI Meeting Tool for Financial Advisors

Nitrogen has launched its Q3 2025 product release, featuring AI Meeting Center and upgrades to other tools. AI Meeting Center is a meeting assistant that creates meeting notes automatically and integrates with platforms like Zoom and Microsoft. Other updates include Firm Controls for firm-wide oversight, Risk Center with asset class drill-downs, Planning Center with a flexible interface, and Research Center with an allocation optimizer. These updates aim to help advisors provide personalized advice and improve client relationships.

Topeka Schools Use AI to Help Teachers

Topeka Public Schools is using artificial intelligence to help teachers create lesson plans and meet individual student needs. The district sees AI as a tool to support teachers, not replace them. While students have limited access, teachers can use AI to personalize learning. The district believes AI is a natural step in the evolution of education, similar to the introduction of microwaves and cell phones.

UK Must Act Now on AI Chip Design Says Report

A report urges the UK to invest in its own AI chip design industry to avoid relying on other countries for this important technology. The report says the UK needs to train more chip designers and create a coordinated plan between government departments. The goal is for UK companies to design 50 new AI chips in the next five years. Experts say the UK has strong research but needs to improve access to design tools and licenses.

Pennsylvania Expands AI Use in Government Agencies

Pennsylvania is expanding the use of AI tools in government agencies to help employees with routine tasks. A pilot program using ChatGPT Enterprise showed that employees saved time on writing, research, and summarizing information. The state is exploring ways to give more employees access to AI tools, with training on safe and responsible use. While AI can help, experts warn that it can also make mistakes, so its work must be checked. Allegheny County is developing its own AI policy.

AI Mental Health Tools Pose Risks, Require Ethical Safeguards

AI mental health tools can help people but also pose risks like manipulation and psychological harm. Studies show a link between AI chatbot use and increased depression. Experts say regulations are needed to ensure transparency and human oversight. Investors should prioritize companies that use ethical safeguards and avoid those that don't. The future of AI in mental health depends on balancing innovation with user safety.

Robinhood Launches AI Investing Tool in the UK

Robinhood has launched Digests by Robinhood Cortex, an AI-powered tool, in the UK. This tool helps investors understand stock movements by providing clear explanations of price changes. Digests uses breaking news, analyst reports, and other data to offer insights. The tool is free for all UK customers and is designed to help investors make informed decisions. It has already been used by many customers in the US with positive feedback.

Fake Hurricane Photos Spread Online Warns News Station

As Hurricane Erin approaches North Carolina, fake photos and videos generated by AI are spreading online. These fake images are designed to create confusion and drive traffic to certain websites. Governor Josh Stein is warning people to get their information from trusted sources like local news and the National Weather Service. There are ways to spot fake images, such as looking for abnormalities and doing a reverse image search.

Sources

• The AI security crisis no one is preparing for
• Lenovo chatbot breach highlights AI security blind spots in customer-facing systems
• Lenovo AI Chatbot Flaw Allows Remote Script Execution on Corporate Systems
• Google Cloud Unleashes Latest AI Security Capabilities
• Google Cloud unveils AI ally for security teams
• Elementary education, artificial intelligence among new RSU degree programs
• Nitrogen Unveils AI Meeting Center, Delivers Free Upgrades to Advisors Using Investment Research & Proposal Generation Products
• Topeka Public Schools implements artificial intelligence in the classroom
• UK urged to seize 'once-in-20-years' AI chip design opportunity
• Pennsylvania government agencies aim to expand employees’ use of artificial intelligence
• AI and Mental Health: Navigating Liability, Ethics, and Investment Risks in a Rapidly Evolving Landscape
• AI investing tool Digests by Robinhood lands in UK
• 5 On Your Side warns against AI-generated Hurricane Erin photos