China team EvoMap erupts in anger over Hermes Agent copying amid viral success: the self-evolving system has a high degree of similarity

ChainNewsAbmedia

A breakout open-source AI agent project, Hermes Agent, has recently become embroiled in a plagiarism controversy. In mid-April, the China-based team EvoMap released a long statement, accusing the self-evolution framework of Hermes Agent—under Nous Research—of having “high similarity” to the Evolver/GEP architecture it previously disclosed.

(AI also wants Hermes! What is OpenClaw, a self-evolving agent?)

EvoMap slams Hermes Agent for plagiarism

EvoMap believes the core of the controversy is not just similarity in a single function, but also several key modules, including a three-layer memory system; automatic sedimentation of reusable assets after tasks; a periodic reflection loop; on-demand skill loading; and self-improvement—each of which is highly isomorphic to Evolver’s design. More sensitively, in EvoMap’s view, publicly available materials related to Hermes have made no citations or acknowledgments of Evolver, GEP, or EvoMap.

In its statement, EvoMap lays out a timeline. The Evolver repository was made public on February 1, 2026, and by mid-February it had already fully disclosed concepts including the GEP protocol; the three-layer asset system of Gene/Capsule/Event; the reflection loop; and experience sedimentation. By comparison, Hermes Agent’s v0.2.0 version was not officially released until March 12, 2026, while the independent hermes-agent-self-evolution repository was created on March 9, 2026.

EvoMap therefore argues that, at least based on publicly verifiable information, Evolver’s core design was exposed 24 to 39 days earlier than Hermes’s self-evolution module.

Breakout Hermes Agent and behind-the-scenes Nous Research

The named Nous Research is an open-source AI lab that has drawn a lot of attention in the AI and crypto circles in recent months.

(OpenClaw alternative: Build a growing AI assistant with Hermes Agent—full tutorial on pain-free migration of OpenClaw memory)

Nous Research was co-founded by Jeffrey Quesnelle, Karan Malhotra, Teknium, and others. In 2025, it raised a $50 million funding round led by Paradigm, valuing it at $1 billion. Nous’s focus is not a single model, but an entire decentralized, community-driven open-source AI route: on the one hand, it releases the Hermes series of models; on the other, it also lays out distributed training infrastructure such as Psyche Network.

(Nous Research deep dive: a decentralized AI lab with a $1 billion valuation backed by Paradigm—full breakdown of the Hermes models and the Psyche network)

And the Hermes Agent at the center of this storm is one of Nous’s loudest products of late. Hermes Agent positions itself as an “agent that grows with usage.” Its core selling point is a built-in learning loop: it can automatically build skills from task experience, continuously patch skills during subsequent use, preserve persistent memory, and gradually develop a long-term understanding of the user and the project.

The official documentation also explicitly describes it as a built-in self-improving AI agent. Precisely because Hermes has recently gone breakout in the open-source community, this controversy quickly spread from the tech circle to broader AI public discourse.

Chinese netizens rally behind EvoMap, angrily accuse Hermes Agent of plagiarism

Nous Research and the Hermes team have not yet publicly issued a point-by-point rebuttal to EvoMap’s detailed technical architecture comparisons. However, Teknium—one of Nous’s co-founders—has said previously that he had “never heard” of the other team or its project, and claimed the allegations were lies. On the other hand, the incident also quickly escalated due to controversy surrounding the phrase “Delete your account” that appeared in early community responses, along with later actions such as deleting posts and blocking.

As for the community, the sentiment is clearly divided, but overall there appears to be more sympathy toward EvoMap. Chinese media and community posts largely focus on two points: “no acknowledgment” and “architecture-level isomorphism.” Some netizens even describe the incident as an AI-era “code-washing” case. On X, you can also see users citing the timeline, repository structure, and comparison articles, arguing that Hermes should at least positively address whether it was inspired by Evolver.

Of course, there are also more neutral voices saying that “learning from experience” and self-evolving agents are directions that multiple teams may converge on, and you shouldn’t jump to legal conclusions based solely on conceptual similarity.

This article—“EvoMap, the China-based team, angrily accuses the breakout Hermes Agent of plagiarism: high similarity in the self-evolution system”—first appeared in Lian News ABMedia.

Disclaimer: The information on this page may come from third parties and does not represent the views or opinions of Gate. The content displayed on this page is for reference only and does not constitute any financial, investment, or legal advice. Gate does not guarantee the accuracy or completeness of the information and shall not be liable for any losses arising from the use of this information. Virtual asset investments carry high risks and are subject to significant price volatility. You may lose all of your invested principal. Please fully understand the relevant risks and make prudent decisions based on your own financial situation and risk tolerance. For details, please refer to Disclaimer.

Related Articles

Adobe Launches CX Enterprise AI Suite to Automate Corporate Marketing with OpenAI, Anthropic Partnerships

Adobe has introduced CX Enterprise, an AI suite aimed at automating and personalizing digital marketing for enterprises. It integrates with major platforms and is adopted by leading advertising agencies to enhance customer interactions and improve marketing performance.

GateNews3m ago

Dark Matter Labs Releases and Open-Sources Kimi K2.6 Model

Dark Matter Labs has released the Kimi K2.6 model, boasting enhanced long-context coding and improved autonomous execution abilities. It is now available on multiple platforms for all users.

GateNews43m ago

a16z latest report: Why blockchain is the missing infrastructure piece that AI agents need?

a16z crypto’s latest report says that AI agents are evolving from support tools into economic actors, yet there are still major gaps in core infrastructure such as identity, payments, and cross-platform collaboration. The report emphasizes that as AI becomes involved in governance and transactions, verification mechanisms become the key to trust, and blockchain technology can provide verifiable infrastructure to address these challenges. The future will require cryptographic mechanisms to ensure that AI agents truly represent users’ intent and to change traditional payment systems.

ChainNewsAbmedia8h ago

Moonshot AI Releases Kimi K2.6 with Enhanced Coding and Agent Capabilities

Moonshot AI has released Kimi K2.6, featuring chat and Agent modes on kimi.com. It excels in coding benchmarks, supports 4,000 tool invocations, and upgraded parallel functionality for autonomous scenarios.

GateNews9h ago

Nexchain Smart Actions Brings AI to Autonomous Web3

Nexchain introduced Smart Actions, a suite of intelligent modules designed to transform blockchain networks from manual, reactive systems into autonomous and self-optimizing environments, according to an announcement on April 18, 2026. The product reflects the company's commitment to building

CryptoFrontier12h ago

Adobe Expands AI Agent Ecosystem, Partners with OpenAI, AWS, and Payment Giants

Adobe has expanded its Agentic AI ecosystem, launching the CX Enterprise Coworker and partnering with major tech firms to enhance AI-driven enterprise operations, including streamlined payment integration for smoother workflows.

GateNews12h ago
Comment
0/400
No comments