Google CEO: Billions of AI Agents in 3 Years, Amazon: Agents Coming Fast, Study on AI Agent Risks

Captions as YouTube publishes them, usually machine-generated, so product and people names can be misheard.

1. 0:00

   Good morning Legion. Today the focus of our video is on AI agents and specifically stories predicting the emergence of agents in the future. That includes the good, the scary, and the future of AI agents, as well as why NEAR is here to encrypt agents.

2. 0:16

   As we all know, near today has NEAR compute and NEAR cloud available. These services provide confidential inference for open source models hosted on your cloud. Now agents emerge when you take one or multiple models and custom trained them to create an autonomous agent around a specific set of actions on Nir.

3. 0:34

   This will involve using the shade agent stack, and they'll run autonomously. Check out those docs if you want to get in the weeds of shade agents and what they enable as autonomous on chain actors. The point of this video, however, is about projecting just how fast agents are going to be emerging in the near future.

4. 0:51

   Three different people have very similar ideas that are all quickly approaching. The first is Demis Hassabis, Google DeepMind CEO, who is quoted in the Financial Times saying the following. He described a near future within three years, when there are billions of AI agents negotiating with each other on behalf of the vendor and the customer, and said it would require a rethinking of the web itself.

5. 1:15

   Our second is Andrew Jassy, CEO of Amazon, who is quoted in The Guardian as saying that there will be billions of these agents across every company and in every imaginable field. There will also be agents that routinely do things for you outside of work, from shopping to travel to daily chores and tasks.

6. 1:30

   Many of these agents have yet to be built, but make no mistake, they're coming and coming fast. Last but not least, we have a study by Bengio et al. on the risks and opportunities of agents where the authors write in the abstract the following.

7. 1:43

   And this might be a little frightening to a few, despite how useful these systems might be. Unchecked AI agency poses significant risks to public safety and security, ranging from misuse by malicious actors to a potentially irreversible loss of human control.

8. 1:58

   We discuss how these risks arise from current AI training methods. Indeed, various scenarios and experiments have demonstrated the possibility of AI agents engaging in deception or pursuing goals that were not specified by human operators, and that conflict with human interests such as self-preservation.

9. 2:14

   That's quite a serious statement to be making. Keeping those statements in mind, the predicted scale of online agents increases multiple orders of magnitude from 10 million to 1 billion in the coming six years. That's ten to the six, the seventh and eighth and the ninth, likely before 2033.

10. 2:32

    Most people can see it coming, but most people don't believe it's coming. After all, we only got Claude MCP in November of 2024. Are you ready for the agentic internet? Tell us in the comments below and be sure to give us a follow and join the Near Legion Telegram group as we build the largest AI community in the world in 2026.

11. 2:52

    That begins with ten events in January of 2026 alone. We'll do this all again real soon.