AI News & Trends

966 articles • Page 10 of 65

Jun 5, 2026

SafeBreach Labs finds WhatsApp bug hijacks Google Gemini

SafeBreach Labs found that attackers may be able to hijack Google Gemini through a WhatsApp message using a method called indirect prompt injection. This bug lets hidden commands in notifications trick Gemini into following attacker instructions without the user's okay. The issue appears to work with other messaging apps too, and could allow data theft or other dangerous actions. Google says it has updated its defenses and these changes appear to have stopped the exploit. The report suggests this kind of attack may also be a problem for other AI assistants.

Jun 5, 2026

Amazon Expands Trainium Chip Production, Challenges Nvidia's AI Dominance

Amazon, Google, and Meta are now competing over the underlying parts of AI, like special chips, data control, and devices, rather than just the AI models. Amazon is increasing the use of its Trainium chips as an alternative to Nvidia, but overall adoption beyond key partners like Anthropic may still be limited. OpenAI and Anthropic together seem to make up most of the revenue among AI startups, showing a clear lead over others. Other companies like Apple, Meta, and Google are trying new ways to run AI directly on devices, but details are limited. The future of AI competition may depend on who controls the chips, data, and where AI runs, not just the models themselves.

Jun 5, 2026

Apple's Overhauled Siri Launches in September, Uses Google Cloud and Nvidia Chips

Apple plans to launch a new version of Siri in September that uses Google Cloud and Nvidia Blackwell chips for complex tasks. The new Siri may remember past questions, understand what is on-screen, and connect actions across apps. While Apple wants to keep most processing on devices for privacy, some work will go to Google Cloud, which is said to be more powerful for big requests. Privacy protections like data encryption are in place, but experts warn that some risks may remain. This change suggests Apple might be using this setup as a temporary solution until its own AI systems are ready.

Jun 5, 2026

Enterprises Target AI Spending, Route Tasks to Cheaper Models

Enterprises are working to control AI costs by using cheaper models for simple tasks and adding rules to prevent overspending. Finance and engineering teams may disagree over expensive AI usage, but shared oversight and cost checks are becoming more common. Studies suggest that using mid-weight or local AI models can lower costs, and having clear cost dashboards may help teams stay productive. Research appears to show that careful cost monitoring does not always slow down developers, even if it sometimes feels that way. Overall, a balance between managing expenses and maintaining productivity seems to be emerging.

Jun 5, 2026

AI Agent Identity Becomes New Enterprise Security Control Plane

AI agent identity is becoming a key part of enterprise security, acting as a single place to manage and monitor what people, machines, and AI agents can do. Experts say identity may be the best way to control access and respond quickly to problems, since agents might bypass old security boundaries with valid credentials. Research suggests new security designs focus on short-lived identities, constant policy checks, and strong audit trails. Some experts warn the control system itself could be a target, so it should be well-protected. Many companies already use AI agents, and more may adopt unified identity controls to keep risks small and easy to manage.

Jun 5, 2026

Google Antigravity 2.0 expands enterprise AI agents with 1M-token windows

Google announced Antigravity 2.0, a tool that helps companies use AI agents without building their own control systems. Analysts say this is Google's answer to Anthropic Managed Agents, with Google focused on bigger context size and connections to its cloud, while Anthropic aims for more safety. Antigravity 2.0 may allow agents to handle much larger tasks and data, but costs and control issues are not guaranteed to be easy. Experts suggest companies might need to pick between bigger, more flexible tools and safer, more careful ones, and warn about getting locked in to one platform. No official numbers show how many businesses are using these tools so far.

Jun 5, 2026

Microsoft Unveils Scout AI Agents for Microsoft 365, IQ APIs

Microsoft introduced Scout AI agents for Microsoft 365, which may help teams automate tasks through APIs like Context, Tools, and Workspaces. Scout appears to run in Teams and uses Azure for orchestration, connecting securely to company tools and protecting data with Microsoft's security features. Reports suggest Scout records every action and uses policy checks to make sure it follows company rules. Enterprises may use agents for meeting prep, reporting, and bug triage, with KPIs to measure time saved and process efficiency. The system seems to help companies use AI agents while keeping strong security and governance.

Jun 5, 2026

Enterprises Pivot to Multi-Model AI Stacks Amid Rising Costs

Enterprises are facing rising costs as they adopt large language models, which may lead them to use several types of models to manage budgets. Early findings suggest that using smaller models first and only switching to bigger ones when needed could save a lot of money. Vendors are changing their pricing, often offering more flexible plans instead of flat rates, and some may charge only for certain business outcomes. Companies are adding tools to track and limit spending, like dashboards and budget alerts. Mid-sized models and tools that help manage costs may benefit most, while companies using only one model or not tracking spending might face unexpected bills.

Jun 4, 2026

Microsoft Unveils 7 New AI Models, Focuses on Enterprise Governance

Microsoft has announced seven new AI models, including one for efficient coding, and is focusing on making them more useful and less costly for businesses and developers. The company is adding new security and control features across Windows, GitHub, and Azure, which may help companies manage risks. These models are made for specific tasks like reasoning, coding, speech, and images, and can work with other AI services. Microsoft also introduced new tools and policies to help control and monitor how AI is used, which might support following future regulations. The overall plan suggests Microsoft wants to make it easier for companies to use AI safely and efficiently.

Jun 4, 2026

Amazon Expands Trainium AI Chip Use, Challenges Nvidia on Cost

Amazon, Google, and Meta are focusing less on launching big AI models and more on building special chips, better data systems, and ways to run AI on different devices. Amazon appears to be using its Trainium chips more, which may help with costs compared to Nvidia, especially for certain kinds of AI work. Microsoft seems to be making its data tools easier for businesses but may also make it harder for companies to switch away from Microsoft later. Reports suggest most revenue in the AI space is now going to just a couple of companies, like Anthropic and OpenAI, though the exact numbers may vary. This situation means businesses might have to consider hardware, data platforms, and costs together when choosing AI tools.

Jun 4, 2026

Enterprise AI spending shifts to infrastructure, hitting $2 trillion by 2026

Enterprise AI spending may reach $2 trillion by 2026, with a shift in focus from AI models to infrastructure like hardware and governance. Reports suggest that most companies are investing more in platforms that connect and manage data, policies, and workflows across different systems. Many organizations feel less prepared in areas like infrastructure and talent, which may indicate a long-term need for tools that simplify complexity. Hybrid deployment appears to be growing quickly, and companies view shared infrastructure as a safer way to manage AI. Startups may stand out by offering integrated infrastructure solutions, while CIOs are focusing on solving integration challenges.

Jun 4, 2026

Anthropic's MCP Protocol Integrates OpenAI, Reduces AI Token Use 98.7%

Anthropic's Model Context Protocol (MCP) helps teams connect AI agents to external tools and data and appears to reduce token use by up to 98.7% in some workflows. MCP is being adopted quickly, with thousands of servers reportedly running since late 2024, though exact enterprise adoption rates are not clear. Teams using MCP may see faster development and fewer errors by treating context as versioned code and using layered context files. While there are reports of efficiency gains, detailed outcome studies and peer-reviewed benchmarks are still limited. Experts suggest that wider and clearer reporting will help determine where context engineering works best.

Jun 4, 2026

OpenAI uses Codex AI to migrate 600 petabytes in two months

OpenAI reportedly used Codex AI to help move 600 petabytes of data in two months in early 2025. Engineers used Codex agents to break down the migration into many small, checkable steps, which may have saved months of manual work. The process still needed humans to supply missing details, supervise, and approve actions, since the AI could miss important context or make mistakes. Reports suggest this hybrid approach lets teams automate more, but there may still be challenges like code errors or cost overruns if teams rely only on the AI.

Jun 4, 2026

Corporations Curb AI Use as Soaring Costs Shift Focus to ROI

Corporations appear to be cutting back on AI use because rising costs are making it harder to justify spending. Companies now focus more on proving clear returns before starting new AI projects, especially in areas where gains are easy to measure, like coding help and customer support. Practical steps, such as using smaller models and setting spending limits, may help lower costs without harming performance. Some firms are choosing different technical setups and vendor options to better control expenses, though these choices can come with trade-offs. Experts suggest that proving value and closely tracking spending might become the standard way for companies to manage AI going forward.

Jun 4, 2026

Nvidia acquires Kumo AI for $400M, expands enterprise software play

Nvidia has acquired Kumo AI, a five-year-old enterprise software startup, for at least $400 million, according to people familiar with the deal. The purchase may help Nvidia expand its software offerings, especially in making predictions for things like fraud detection and demand forecasting. Kumo's technology appears to quickly turn raw business data into predictions, and it claims to work with companies like Reddit and DoorDash. Some analysts say this move fits into Nvidia's recent pattern of buying AI software firms, suggesting a strategy to be more central in business workflows. However, performance claims mainly come from the vendor, and some details about the deal have not been made public.