AI agents are moving beyond chatbots to handle complex, multi-step tasks. Major tech companies are integrating these tools into desktop and business environments. The shift raises new questions about access, authority and control.
Media and publishing businesses are facing a new wave of AI integration as leading tech companies roll out agentic tools capable of executing complex, multi-step tasks across files, apps and workspaces. This evolution from simple chatbots to advanced AI agents is prompting urgent questions about how much access and authority these systems should have inside editorial and media-planning environments.
Google this week launched "Gemini Spark" for macOS, allowing users to assign tasks from their phones-such as retrieving specific reports, calculating revenue totals and emailing results-while the agent works autonomously on their desktop. The beta is currently available to Google AI Ultra subscribers in the U.S. The move positions Google to compete more directly with desktop AI agents like Claude Desktop, Microsoft’s Copilot and OpenClaw, as Gemini Spark gains the ability to interact with local files and soon, remote tasks. Google is also introducing support for custom Model Context Protocol (MCP), enabling direct integration with users’ preferred apps.
OpenAI is also shifting its focus from chatbots to agentic AI. According to executives, the company aims to transform its chatbot into a “superapp” that combines coding tools and AI agents, with the goal of generating more revenue from products than from ads or subscriptions. OpenAI’s strategy is already visible in updates to ChatGPT’s website and mobile apps, which now encourage use of coding, image-generation and third-party apps. Executives say these changes are designed to make agentic AI more valuable for business users, moving closer to Anthropic’s approach of targeting professional workflows.
Anthropic recently launched "Claude Sonnet 5," described as its most agentic model to date. The new version can plan, use tools like browsers and terminals, and run autonomously at a level that previously required larger, more expensive models. Early testers report that Claude Sonnet 5 completes complex tasks and checks its own output, where earlier models would stop short. Amazon Web Services has made Claude Sonnet 5 available on Amazon Bedrock, allowing developers to build with it inside AWS environments while maintaining enterprise security and data residency.
The rapid adoption of agentic AI is already influencing how publishers and platforms adapt their operations. For example, some major publishers are redesigning their websites to remain visible to AI-driven search and agentic tools, as reported in coverage of Time and The Economist’s AI-optimized site formats.
As agentic technology becomes more capable, concerns are growing about the privileges these systems hold. AI agents can authenticate, receive permissions, call APIs, write code, trigger workflows, query databases and act across production environments. The question of whether AI needs a “kill switch” is becoming more urgent as these tools gain deeper access to business-critical systems.
Alphabet, Google’s parent company, was recently added to the Dow Jones Industrial Average, reflecting its growing influence in AI. In late June 2026, Adobe was included in several Russell value and defensive benchmarks, highlighting the increasing role of AI-focused companies in major stock indexes.
