Get AI Briefings, Transcripts & Alerts on Local & National Government Meetings — Forever.
VBCPS presents phased AI plan; board presses for clearer middle‑school guardrails and monitoring details
Loading...
Summary
Director Ryan Goldberg outlined a phased AI implementation — high‑school access to Google Gemini, no generative models for elementary, and a planned middle‑school rollout using walled‑garden platforms — while board members pressed for stronger guardrails, monitoring cadence and clarity on vendor vetting and policy control.
The Virginia Beach City Public Schools board received a detailed update April 28 on the division's artificial intelligence strategy, training and usage data, prompting several board members to urge caution about introducing AI to middle‑school students.
Ryan Goldberg, director of instructional technology, said VBCPS has emphasized "selective and intentional" AI use aligned to instructional goals and has provided professional learning for staff. He noted high‑school students were granted access to Google Gemini on Nov. 5 and cited engagement figures: 69% of high‑school students have used Gemini to some degree, with about 46% classified as light users, 21% regular users and 2% "power users." Goldberg stressed that AI‑generated grades are prohibited and that teachers remain responsible for evaluating student work.
"Our focus remains on sound instruction, ensuring that technology serves to spark creative thinking rather than reduce critical thinking opportunities," Goldberg said.
On monitoring and safety, Goldberg described a newly released Securly chat‑monitoring feature that allows authorized staff to view student prompts and AI responses; he said the division plans to train staff and not enable the feature until the 2026–27 school year to ensure proper use. A division IT/security representative said Securly generates alerts that trigger human review when content suggests a risk of imminent harm.
Board members pressed for clear, grade‑level guardrails. Vice Chair Miss Rogers, identifying herself as an early‑childhood educator, asked for specifics about the guardrails for each level and expressed concern about the 2% of power users at the high‑school level and whether similar or greater misuse could occur among younger students. Goldberg replied that elementary and middle school students will not access generative models; the administration plans a middle‑school pilot with a walled‑garden instructional platform that provides teacher dashboards and alerting rather than open generative models such as Gemini.
"Middle school will not have access to Gemini," Goldberg said. "We are thinking about recommending some form of an instructional platform'1 that is more secure and offers teacher oversight."
Several board members asked about vendor vetting, data privacy and the use of free tools. Goldberg and technical staff replied that free tools can be used only with worksheets that contain no student identifiable information; tools that handle sensitive data require a digital privacy agreement (DPA) or other contractual protections before regular classroom use.
Questions about accuracy and "hallucinations" also drew sustained discussion. Goldberg said the division trains teachers to vet AI outputs and to treat AI products as instructional aids that require teacher review before use in class.
Board members asked what process would be required to stop or delay a middle‑school rollout. Policy staff said the board could adopt a resolution or revise the policy to restrict implementation; staff indicated the plan would return to the board at the retreat and through policy review for additional board input.
The board recessed for a break and asked Goldberg to return to complete remaining questions at the formal meeting.
Next steps: staff will continue vetting instructional platforms for middle school, provide additional stakeholder summaries and return with recommendations and policy options for board consideration.
