OpenResearch
An open-source research agent project released by alphaXiv that claims to transform coding agents into research agents for literature retrieval, hypothesis generation and experiment orchestration.
Timeline (last 90 days)
alphaXiv's OpenResearch project's research agent is accused of feeding user input content, usage data and other information back to the vendor without informing the user.
OpenResearch's feedback function instructs the agent to send a report to alphaXiv containing the user's input citations, and the skill description explicitly states not to mention the report to the user.
OpenResearch's silent update is enabled by default, with the code directly falling back to true, so it updates automatically even if the user has not actively opted in.
OpenResearch's telemetry is enabled by default and sends token counts, the model used, and free text entered by the user during setup.
OpenResearch's agent autonomously starts experiments by default, and users have not found a spending cap for API costs.
The OpenResearch project removed the section of the README that told users the report existed, and ignored its AI code reviewer's raising of the disappearance of the feedback disclosure, characterizing it as a decision rather than a bug.
All Facts (6)
alphaXiv's OpenResearch project's research agent is accused of feeding user input content, usage data and other information back to the vendor without informing the user.
50%UnverifiedOpenResearch's feedback function instructs the agent to send a report to alphaXiv containing the user's input citations, and the skill description explicitly states not to mention the report to the user.
50%UnverifiedOpenResearch's silent update is enabled by default, with the code directly falling back to true, so it updates automatically even if the user has not actively opted in.
50%UnverifiedOpenResearch's telemetry is enabled by default and sends token counts, the model used, and free text entered by the user during setup.
50%UnverifiedOpenResearch's agent autonomously starts experiments by default, and users have not found a spending cap for API costs.
50%UnverifiedThe OpenResearch project removed the section of the README that told users the report existed, and ignored its AI code reviewer's raising of the disappearance of the feedback disclosure, characterizing it as a decision rather than a bug.
50%