• by Laurie Sullivan , Staff Writer, August 19, 2025

When Bill Gates founded Microsoft, he envisioned more than a software company. He envisioned a “software factory, unconstrained by any single product or category.”

That idea guided the company for decades, but today it’s no longer enough, Satya Nadella, Microsoft chairman and CEO, wrote in a note to employees in July.

Since that letter, AI's pervasive integration across Microsoft's applications and services -- from advertising to gaming and Excel -- while presenting numerous opportunities also poses several challenges for advertisers that extend beyond the ads they make. Sometimes, ads and images can be turned into deepfakes.

AI agents also have slowly crept into the settings of its PCs. When a consumer needs to change the settings on their PC, the Windows new agent in Settings and Copilot Vision features on Windows makes it for the user. It will fix audio levels or connect to a printer.

Agent in Settings uses on-device AI to understand the user’s intent, and with permission, automates and executes tasks. The user can describe what’s needed -- for example, “I want to control my PC by voice” or “my mouse pointer is too small” -- and the agent in Settings will recommend the steps to resolve it.

Charles Lamanna, corporate vice president of business and industry for Copilot at Microsoft, earlier this year likened current systems to legacy mainframes, and believes the future belongs to AI agents that do not follow rigid workflows or forms, but instead adapt to users and goals.

The growing presence of AI agents like Copilot and features like Windows Recall within Windows settings presents opportunities as well as significant challenges for advertisers and software developers.

Integrating AI agents into existing advertising workflows can become complex. AI algorithms can reflect biases present in the training data, leading to discriminatory targeting or content. Advertisers have a responsibility to address this by diversifying training data, monitoring AI outputs, and implementing ethical guidelines.

And while research suggests consumers are receptive to ads in conversational AI, concerns about privacy and ad relevance could impact trust. Advertisers need to communicate to consumers which type of data they can see. 

AI models in Windows settings, for example, can access and analyze user data, including personal messages, search history and potentially sensitive information within documents and applications. 

Microsoft integrated various AI agents in its advertising platforms earlier on. Copilot is the central AI assistant within the platform, and is aimed at empowering advertisers with generative AI capabilities. It supports all types of tasks from campaign setup to performance optimization in platforms like Showroom Ads or Copilot Studio.

Although it is not specific to Microsoft, 87% of game developers already use AI agents, according to a survey from Google Cloud and The Harris Poll of 615 game developers in the United States, South Korea, Norway, Finland, and Sweden.

The data suggests that AI is solving problems within the game, increases speed of the prototype, improves quality controls and testing, and supports brainstorming about the storyline.

There is also a downside to AI and agents. All types of deepfakes are becoming a challenge -- such as CEO impersonators, according to Brian Long, chief executive and co-founder of OpenAI-backed cybersecurity firm Adaptive Security. Cyber crooks can make one video from one image along with seconds of audio.

Cyber criminals are using deepfake voice and videos of top executives to take companies from millions of dollars, according to the Wall Street Journal, but they also are creating content to use as advertisements.

Reportedly, Nexon, a video game company behind the video game "The First Descendant," used a content creator’s likeness to promote the game.

The streamer, DanieltheDemon, claimed Nexon used his face and reactions from his most viral video without his consent, altering his mouth movements and voice with AI. The ad appeared on TikTok with several “irregularities" in the ads, which were submitted by users through a creative challenge.

TikTok's system was supposed to verify the content for copyright issues before being used as ads.