Texas issues AI training requirements and adopts DIR AI rules; sandboxes coming this summer

Jenny Holster of the Texas Department of Information Resources briefed LTAC on the agencys progress implementing AI policy and training.

Holster said HB3512's training requirement (added to the existing cybersecurity training statute) went into effect Sept. 1 and DIR has begun certifying training programs. "We have 13 that have been already certified," she said, and those certified programs will be listed on DIRs website. The training requirement applies to the same staff who take the state cybersecurity training, she added.

The DIR board adopted agency AI rules in a public meeting; Holster said the rules were sent to the secretary of state and will be published and become effective roughly 20 days after adoption (she estimated an effective date around March 18). Holster also described an AI code of ethics and minimum standards for heightened-scrutiny AI tied to Senate Bill 1964.

Holster described legislative direction for DIR to create AI sandboxes where state agencies and institutions can test AI systems in a controlled environment. She said the sandboxes are under development and that she expects them to be available in the summer, with outreach to information-resource managers and the DIR AI user group used to invite participants.

Why it matters: The new training requirement and DIR rules establish a baseline regulatory and ethical framework for AI use in Texas government and public higher education institutions. Sandboxes could offer institutions a low-risk way to pilot tools ahead of broader deployment.

Next steps: Holster said she would send LTAC staff a link to certified training programs and provide updates when the DIR rules are published; members were encouraged to consult local IT and legal teams about training timelines and participation in the sandbox pilot.