Sentient and EigenLayer Launch AI Judge That Doesn't Make Any Decisions
On June 11, Sentient and EigenLayer announced the launch of Judge Dobby — an AI system for analyzing corporate disputes. The press release positions it as the "first AI adjudicator" for work in "high-stakes areas," including corporate governance and public resource allocation.
According to the release, Judge Dobby "embeds accountability into AI-driven dispute resolution" and "ensures decisions can be trusted and traced."
But responsibility falls on the "existing stakeholders"
When I asked who bears legal and financial responsibility when Judge Dobby makes an incorrect decision or gets compromised, company representatives provided the following answer: "Judge Dobby doesn't make any decisions. It's one data point of many for the community to consider before voting in accordance with established DAO rules." They attributed this response to an Eigen Labs spokesperson.
When asked about whether protocol users simply bear the financial losses with no recourse, the response was: "Judge Dobby will be a tool for protocol users to refer to, like other tools as part of their research in deciding for or against a proposal. As such, the responsibility still falls on the existing stakeholders."
When asked if any insurance, appeals processes, or compensation mechanisms exist, the response was: “As Judge Dobby will not be making any decisions, no.”
It remains a mystery to me how the same system can simultaneously "automate protocol, community, and governance compliance across its ecosystem," while also not making any decisions.
And they chose not to answer my questions
According to the press release shared with us by the company's PR representative, titled "Sentient Launches First-of-Its-Kind AI for Adjudicating Corporate Governance Disputes," Judge Dobby was presented as a system capable of handling corporate disputes where "objectivity fails."
After receiving their initial responses about responsibility, I asked about a specific contradiction: the press release states that Judge Dobby "embeds accountability into AI-driven dispute resolution" and "ensures decisions can be trusted and traced," yet their responses showed that the system doesn't make decisions and no one bears responsibility for its recommendations.
The PR representative then requested "assurance that the piece is going to be balanced rather than going in a completely negative angle" to "help reassure the team."
When I explained that I cannot pre-commit to any particular angle, they left my questions unanswered and instead clarified: "This is a research-stage announcement, not a product launch. The Judge Dobby system is still in early development and has not been finalized for production."
The PR representative also added that "many of the important accountability questions you've raised are exactly the issues this research seeks to explore further," even though the press release announced a "launch" without any mention that Judge Dobby was experimental or that accountability mechanisms were still "open areas of inquiry."
Questions left unanswered
The questions that received no response included how Judge Dobby "embeds accountability" if no one bears responsibility for its recommendations, whether the system prompt will be published for transparency, what mechanisms prevent bias favoring certain parties, who approved the algorithm for "high-stakes" use, and whether independent audits were conducted.
They also did not respond to my request to identify their spokesperson beyond "Eigen Labs spokesperson" for proper attribution.
