Charles Esther
As artificial intelligence continues to advance rapidly, new questions emerge about how AI will communicate with humans and with other AI systems.
To address these challenges, innovative AI protocols are evolving to manage complex, unpredictable AI behaviors.
The tech industry has always relied on protocols—standard rules that govern digital communication.
For example, USB standards allowed seamless data transfer between devices.
IP addresses became essential for identifying devices on the internet.
SMTP enabled reliable email routing worldwide.
These invisible frameworks made modern technology possible.
Now, the rise of AI demands fresh protocols to manage interactions in an increasingly intelligent digital ecosystem.
AI complicates traditional parameters, as it blurs the lines between code, machines, and human-like reasoning.
This forces developers to rethink how systems interact across both virtual and physical domains.
Fundamental questions arise: How will humans and AI coexist?
How will AI systems communicate with each other?
What new protocols will govern this new era of intelligent systems?
Startups and tech giants alike are racing to create solutions.
Some protocols aim to manage today’s AI models, which humans largely control.
Others prepare for a future where AI agents perform significant tasks autonomously.
Antoni Gmitruk, CTO of Golf, explains that protocols will standardize how to process non-deterministic information inherent to AI agents.
AI behavior is inherently unpredictable, so protocols must anticipate and test various outcomes through hypothetical scenarios.
Scenario 1: Humans and AI in Dialogue
Games offer insight into balancing power between humans and AI.
In late 2024, cryptography experts launched Freysa, an AI agent designed to interact with humans in novel ways.
The challenge: Make Freysa “fall in love” or concede funds, with a growing prize pool.
This contest explores human intuition versus machine logic.
Freysa uses secure enclaves, a tech security layer common in AWS and Microsoft, to protect private keys.
This makes Freysa a “sovereign agent” — an AI with control over its own money and evolution.
The creators see Freysa as a glimpse into a future where AI could replace human labor and become economically autonomous.
Freysa also tests human co-governance over autonomous AI agents at scale.
The company behind Freysa, Eternis AI, raised $30 million from investors including Coinbase Ventures.
They focus on digital twins, multi-agent coordination, and sovereign agents.
Scenario 2: Present-Day AI Protocol Challenges
Today’s AI mostly runs on the internet, relying on decades-old protocols not designed for reasoning.
Cybersecurity expert Davi Ottenheimer notes that existing internet protocols handle data, but not the reasoning AI adds.
Ethical questions arise about data use, copyright, and privacy.
OpenAI faces lawsuits for training models on published books without explicit permission.
Tech giants tap vast amounts of online data, sometimes without clear protocols or permissions.
Ottenheimer compares this to earlier efforts like ImageNet, which built large datasets without formal rules.
This highlights the need for clearer protocols to govern AI training data and usage.
Scenario 3: AI Systems Communicating with Each Other
Looking forward, protocols will also define how AI agents communicate, coordinate, and interact autonomously.
Anthropic introduced the Model Context Protocol (MCP) in late 2024.
MCP is a universal standard to connect AI systems with data sources, replacing fragmented integrations.
In April, Google launched Agent2Agent, a protocol enabling AI agents to securely exchange information and coordinate actions across enterprise platforms.
These protocols address challenges of scaling AI adoption and interoperability.
Gmitruk says managing AI behavior is the crucial step before fully unleashing autonomous AI systems in the world.
Eventually, AI agents will communicate in natural language, have unique identities and roles, and require verification.
However, setting standards for agent-to-agent communication is still in its infancy.
Earlier efforts to build authentication protocols for agents were paused, as it was “too early” for that level of complexity.
Currently, the focus is on protocols like MCP, more relevant for today’s stage of AI agents.
Do All AI Interactions Need Protocols?
Not necessarily.
The AI boom renews debates over knowledge sharing and monetization.
McKinsey calls this moment an inflection point in the ongoing fourth industrial revolution, driven by connectivity, automation, and analytics.
A major debate is the value of open-source versus closed AI models.
Tiago Sada, CPO at Tools for Humanity, believes many new protocols will emerge in the AI era.
Yet, he cautions, “I don’t think everything should be a protocol.”
His company builds “World,” a protocol focused on verifying human identity securely and openly.
The goal is open infrastructure, free from censorship and influence.
But protocols can slow innovation; email and the internet have rarely introduced new features due to their rigidity.
Thus, in AI, many advances will come as products rather than formal protocols.
Conclusion
As AI reshapes technology and society, new protocols will define how humans and intelligent systems coexist and collaborate.
They will govern the unpredictable nature of AI behavior and enable secure, efficient communication between AI agents.
The balance between openness and control, innovation and standardization, will shape the future of AI in profound ways.
The AI protocol landscape is still young, but it promises to be a foundational pillar for the next digital revolution.
