- Covenant AI said it is leaving Bittensor because the network’s governance no longer reflects its stated commitment to decentralization.
- Founder Sam Dare said the split was driven by a loss of trust in how power is exercised inside the ecosystem.
Covenant AI is leaving Bittensor, and the company is framing the decision as a break over principle rather than a routine ecosystem dispute.
In a statement published on X, founder Sam Dare said Covenant can no longer continue building on Bittensor because the network’s governance structure contradicts the decentralization it publicly claims to uphold. The language was sharp and plainly intentional.
Dare said the promise that no single entity controls the network was the core idea that brought builders, miners, validators and investors into the ecosystem, then added that this promise had proved false.
Covenant says the core premise no longer holds
That is a serious accusation, especially coming from one of the better-known teams on the network. Covenant said it had spent its time on Bittensor working from the belief that AI model training should not sit under the control of a single party.
The team pointed to Covenant-72B, which it described as the largest decentralized instance of LLM pre-training ever run, as proof that it had followed through on that vision in practice.
That work also made Covenant one of the most visible and important subnets in the Bittensor ecosystem. So this is not a fringe project walking away quietly. It is a prominent builder saying the governance reality no longer matches the founding narrative.
The departure centers on control, not product failure
Dare identified Bittensor co-founder Jacob Steeves, also known as Const, as the central reason for the departure. According to the statement, Steeves had been asserting authority over the subnet in an effort to regain control of Covenant after it had grown beyond what the team believed could be centrally managed.
That shifts the focus of the story away from token price reactions and back to the more consequential issue. Covenant is not leaving because its product failed or because the market turned. It is leaving because it no longer believes Bittensor operates according to the decentralization standard that justified building there in the first place.
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
Claude Mythos: Does it pose a threat to financial security? The U.S. Treasury Secretary and the Chair of the Federal Reserve hold an emergency meeting to warn about risks
The U.S. Treasury Secretary and the Federal Reserve Chair convened top executives on Wall Street, warning that the AI model Mythos could pose systemic risk to the financial system. Mythos can autonomously discover large numbers of vulnerabilities; Anthropic chose to limit its public release and launched the “Glass Wing Program” to strengthen cybersecurity.
CryptoCity2h ago
Anthropic hires lobbying firm Ballard Partners, after talks with the Pentagon fell apart over AI use restrictions
Anthropic announced on April 14 that it is partnering with the lobbying firm Ballard Partners, aiming to improve its relationship with the Department of Defense. The two sides had previously failed to reach an agreement due to disagreements over the scope of AI use: the Department of Defense wanted its tools to be used without restrictions, while Anthropic demanded limits on their use for fully autonomous weapons and surveillance. This move signals a significant increase in policy outreach investment in the AI industry.
GateNews2h ago
OpenAI internal memo leaked: directly calls out Anthropic’s fear-based narrative and “compute is behind, revenue is inflated”
Artificial intelligence competition is intensifying. In an internal memo, OpenAI emphasized the need to strengthen its enterprise market, increase user stickiness, and criticized rival Anthropic’s strategic missteps and exaggerated financial figures, pointing out its disadvantages in both compute power and product strategy. OpenAI plans to evolve into a multi-product integrated platform, using its compute advantage to gain market leadership.
ChainNewsAbmedia3h ago
AI medical care exposes discrimination! High-income patients receive precise testing, while Black patients and people experiencing homelessness are recommended for invasive treatment.
Recent research indicates that medical AI provides different medical recommendations to patients based on their income, race, and other backgrounds. As a result, higher-income individuals are recommended higher-end tests, while disadvantaged groups are more likely to face improper medical treatment, which could worsen bias and increase the risk of clinical misdiagnosis. Experts are calling for the creation of monitoring mechanisms to ensure healthcare equity.
ChainNewsAbmedia3h ago
Investors Question OpenAI’s $852 billion Valuation, Saying a Strategic Shift May Face Competitive Threats
Gate News message. On April 14, the Financial Times reported that OpenAI investors questioned its $852 billion valuation, saying the company’s strategy is shifting. Some investors said these strategic changes may make OpenAI more vulnerable to threats from competitors such as Anthropic and Google.
GateNews4h ago
Boss, you don’t need to go into the office anymore! Meta is building Mark Zuckerberg’s digital twin—if you have anything to ask, just ask the AI.
Meta is developing an AI clone based on its CEO, Mark Zuckerberg, to strengthen connections between employees and management. This digital avatar will use Zuckerberg’s image, voice, and strategic thinking, with the goal of enabling interaction for about 80k employees. This initiative is different from the “CEO agent,” which is a decision-support tool. Meta’s long-term goal is to develop “superintelligent” AI and to try applying this technology to influencers and creators around the world.
ChainNewsAbmedia4h ago
