The EU is banning using AI-powered digital assistants throughout on-line conferences. Such assistants are sometimes used to transcribe, take notes, and even file visuals and audio throughout a video convention.
In a presentation from the European Fee delivered to European Digital Innovation Hubs earlier this month, there’s a word on the “On-line Assembly Etiquette” slide that states “No AI Brokers are allowed.”
AI brokers are instruments that may carry out complicated, multi-step duties autonomously typically by interacting with purposes, reminiscent of video conferencing software program. For instance, Salesforce makes use of AI brokers to name gross sales leads.
The Fee confirmed this presentation was the primary time this rule had been imposed however declined to elucidate why when questioned by Politico. There is no such thing as a particular EU laws that covers AI brokers, however the AI fashions that energy them might want to abide by the strict and controversial guidelines of the AI Act.
AI brokers elevate safety considerations
Whereas AI notetakers and different agent sorts should not inherently a safety risk, based on a 2025 report from world AI consultants, safety dangers stem from the person being unaware of what their AI brokers are doing, their innate skill to function exterior of the person’s management, and potential AI-to-AI interactions. These components make AI brokers much less predictable than commonplace fashions.
SEE: How Can AI Be Used Safely? Researchers From Harvard, MIT, IBM & Microsoft Weigh In
Tech corporations do should be cautious when selling merchandise that may accomplish an growing quantity with out the person’s consciousness. One of many largest cautionary tales is that of Microsoft Recall, an AI instrument that allowed customers to manage their PC or search by way of recordsdata utilizing pure language.
The comfort got here at a price: Recall captured screenshots of energetic home windows each few seconds, saving them as a timeline, elevating considerations about privateness and information utilization and resulting in vital launch delays. Microsoft has since launched a sequence of brokers particularly designed to deal with cyber threats.
AI brokers are rising in prevalence
This hasn’t stopped the AI gamers from handing over extra management to their fashions. Anthropic added a Pc Use characteristic to its Claude Sonnet chatbot in October 2024, which gave it the power to navigate desktop apps, transfer cursors, click on buttons, and sort textual content. Its deep analysis perform, introduced this week, additionally responds to prompts “agentically,” as does Microsoft’s equal.
Final month, OpenAI expanded its text-to-speech and speech-to-text instruments to agentic fashions, indicating their rising relevancy. In January 2025, OpenAI introduced Operator, an agentic instrument that runs in-browser to autonomously carry out actions reminiscent of ordering groceries or reserving excursions.
SEE: EU Invests €1.3 Billion to Increase AI Adoption & Enhance ‘Digital Competencies’
Anthropic and OpenAI are even working collectively to enhance agent know-how, with the latter including help for the previous’s Mannequin Context Protocol, an open-source commonplace for connecting AI apps, together with brokers, to information repositories. Anthropic has additionally joined forces with Databricks to assist massive company shoppers construct their very own brokers.
TechRepublic predicted on the finish of 2024 that using AI brokers will surge this yr. OpenAI CEO Sam Altman echoed this in a January weblog submit, saying “we might even see the primary AI brokers ‘be part of the workforce’ and materially change the output of corporations.”
By 2028, 33% of enterprise software program purposes will embody agentic AI, up from lower than 1% in 2024, based on Gartner. A fifth of on-line retailer interactions and at the very least 15% of day-to-day work choices will probably be carried out by brokers by that yr.
