Advisory board sets timeline for Connecticut state data plan, flags AI and open‑data deadlines

Scott Gollum, the chief data officer at the Office of Policy and Management, opened the June 18 meeting of the Connecticut Data Analysis Technology Advisory Board with a schedule for the next state data plan and related deliverables.

"I'm required to ... develop a plan, submit it to this board by the end of the calendar year, 12/31/2026," Gollum said, laying out the basic timeline: a draft to the board by Nov. 1, 2026, a public hearing between Nov. 1 and Dec. 31, and a final plan filed by year end.

Why it matters: the two‑year state data plan guides agency work on data inventory, metadata and open access. Board members and agency representatives said the plan will underpin transparency efforts and preparations for new statutory requirements on AI and open data.

Matt Honeau, the state’s new open‑data research analyst, described the status of agency open‑data access plans and a revised schedule. "Each executive branch agency shall, develop an open data access plan," Honeau read from statute and said the project’s publication goal has been pushed to September 2026 after staff turnover delayed earlier work.

Honeau reported three categories of agency responses from prior meetings: agencies with robust, specific goals; agencies whose goals need clarification; and agencies that have not yet submitted plans. He said staff will follow up individually to complete the documentation and that the office is aiming to publish completed plans in September.

The board also discussed legislative items that affect data work, including a recent omnibus AI bill and new transparency reporting on legislatively directed funds (earmarks) that will begin reporting in 2028. Gollum framed the AI changes as paired requirements: agencies must identify data that could support AI systems and develop governance, data‑quality, privacy and transparency policies to make that data safe to use.

Gollum cited an Urban Institute study the board plans to circulate that tested generative AI models on basic government data queries and found inconsistent results; he said that example motivated the requirement to make state data more reliable for automated systems.

Next steps: staff will collect statutory citations and other materials cited during the meeting, complete agency follow‑up on open‑data goals, deliver the draft plan to the board by Nov. 1, and hold a public hearing before finalizing the plan by Dec. 31, 2026. The AI policy and related enterprise work are scheduled to be in place by Jan. 1, 2028.

The board did not take formal votes at the meeting; members were asked to provide materials and statute references by email to staff for follow‑up.