AI literacy is no longer one skill. In April 2026 China ordered a national build-out of full-stage AI education and a society-wide literacy mechanism by 2030. Maryland’s Artificial Intelligence Ready Schools Act, effective June 1, 2026, puts a 120-day policy clock on every local school system and requires AI literacy inside computer science and workforce standards by June 1, 2027. On September 23, 2026, UT San Antonio announced a $500,000 NSF-funded OpenSecCoder platform that treats literacy as critical evaluation of AI-generated code, not trust. Federal H.R.8747 still only unlocks Elementary and Secondary Education Act Title IV funds for safe, effective, and responsible AI instruction. It was ordered reported amended 18–15 on July 21, 2026. It is not law.
The defended position is operational. If your organization trains people on AI, or ships AI that writes code into production paths, “AI literacy” has to mean a named outcome with a deadline and a failure mode. Universal course mandates, state school-policy clocks, and code-distrust labs are three different governance products. Treating them as the same slide deck fails operators who have to staff curricula, adopt local policies, or stop insecure AI-written code from reaching customers.
China’s national mandate: literacy as universal build-out by 2030
On April 2, 2026, China’s Ministry of Education and four other departments issued the “Artificial Intelligence + Education” Action Plan. The plan’s 2030 target is explicit: a vertical-and-horizontal full-stage AI education system plus general education for all of society, with a long-term mechanism for nationwide AI literacy. K–12 AI courses are to move into local curricula, with course guides that set goals, content, and hours by stage. Higher education is told to make AI a public foundation course. Vocational programs must upgrade for intelligent industries. Society-wide literacy resources are to open through national platforms, with customized courses for key groups and credit-bank pathways.
Teacher literacy is not optional language. The plan directs standards for teachers’ AI literacy, layered training with full coverage, situational assessment tools, AI content in teacher qualification exams, and AI in normal-student curricula. Safety is written as a barrier layer: application security systems, model content review, evaluation standards across model, data, infrastructure, and applications, campus product rules, and mechanisms against forgery fraud, academic fraud, exam arms races, and privacy leaks. For operators watching comparative policy, this is literacy as state capacity with a dated deliverable, not a soft skill workshop.
Maryland’s clock: policy in 120 days, literacy in standards by 2027
Maryland Chapter 634 (Senate Bill 720), the Artificial Intelligence Ready Schools Act, took effect June 1, 2026, after gubernatorial approval on May 26, 2026. The State Department of Education must publish AI guidance for local systems, educators, parents, and students that promotes safe, responsible, equitable, and ethical use and promotes AI literacy through computer science education. Within 120 days of that guidance release, every local school system must have an AI policy aligned to it. Each system must designate a noninstructional central-office AI coordinator. The Department must publish a rubric and evaluative tools for AI tool selection; local systems must use them and procure consistent with state AI procurement law.
Literacy is timed. On or before June 1, 2027, the Department, with the Governor’s Workforce Development Board, must ensure AI literacy is a component of workforce preparation standards and computer science standards for kindergarten through grade 12. Professional development must cover purpose, vocabulary, how AI works, classroom use, privacy, security, academic integrity, and avoiding overdependence, with a train-the-trainer intent to reach educators by July 1, 2027. The Maryland AI Education Collaborative on Artificial Intelligence in K–12 Education must deliver curriculum examples by January 1, 2027, and annual reports starting December 1, 2027. This is literacy with a coordinator, a policy deadline, and a standards deadline — Checkpoint-Based Governance with calendar ownership.
OpenSecCoder: literacy as refusing to trust AI-written code
UT San Antonio’s September 23, 2026 announcement describes a third literacy product. Nishant Vishwamitra leads a $500,000 National Science Foundation project to train students to spot and fix security flaws in AI-generated code before they enter the workforce. OpenSecCoder is a free online platform with eight hands-on labs, three course modules, and annual faculty workshops. Students work in a sandboxed interface with step-by-step instructions, a coding agent, a terminal, and a debugger. The project expects to reach about 600 undergraduates and 400 graduate students each year in collaboration with the University at Buffalo. Five of eight labs require no programming background, because AI coding tools are already used in business, health care, and communications.
The literacy claim is explicit and measurable. Students should leave reading AI-generated code critically instead of trusting it, using professional scanning tools, comparing AI tools rather than picking by reputation, and recognizing attacks aimed at agents. Agentic coding tools that create files, install software, run commands, and self-correct without step-by-step human direction are named as a risk surface: agents are built to finish the task, and security is usually what gets sacrificed. Materials will be MIT-licensed so community colleges, minority-serving institutions, and rural universities can adopt them. This is literacy as a verification habit for anyone who ships AI-assisted software, not only cybersecurity majors.
Federal H.R.8747: fund unlock, not a mandate
H.R.8747, the K–12 AI Literacy and Readiness Act of 2026, introduced May 12, 2026 by Rep. Fine, amends the Elementary and Secondary Education Act so Title IV state and local allowable uses cover instruction on using AI safely, effectively, and responsibly, plus professional development for teachers, paraprofessionals, librarians, support personnel, and administrators. On July 21, 2026, the House Education and Workforce Committee ordered the bill favorably reported to the House as amended by a recorded vote of 18 yeas and 15 nays. That is committee progress. It is not enacted law, and it does not impose a national literacy curriculum, a 120-day local policy clock, or a code-security training requirement. Compared with China’s dated build-out and Maryland’s timed standards, the federal bill is an eligibility unlock for existing ESEA dollars.
What Responsible AI operators should change now
The Factics move treats the China Action Plan, Maryland Chapter 634, the H.R.8747 markup, and the UTSA OpenSecCoder announcement as verified primary events. The tactic is to stop using one “AI literacy” label for three jobs. Map every training, school-policy, or engineering program to one of three outcomes: (1) broad population fluency with course hours and teacher standards, (2) institutional policy and coordinator ownership on a calendar, or (3) critical evaluation of AI-generated artifacts before production use. The KPI within one reporting cycle is the share of AI-facing roles with a named literacy outcome, an owner, a deadline, and a failure test — for example, can the trainee reject insecure AI-written code, or can the school show an adopted policy within 120 days of state guidance.
Checkpoint-Based Governance asks who owns the literacy gate before AI tools reach students or production systems. A workshop attendance sheet is not a checkpoint. A dated policy, a designated coordinator, or a lab that forces students to break AI-written code is. Factics keeps the three literacy products separate so KPIs do not collapse into vanity completion rates. HAIA-CORE is the evaluation method that forces those claims into readable structure for boards comparing national mandates, state clocks, and federal fund unlocks on the same evidence.
Watch the next enforcement layer. Maryland’s Collaborative reports and CS/workforce standard updates will show whether the 2027 literacy deadline is real. OpenSecCoder’s MIT release will show whether critical code evaluation scales beyond one NSF project. Federal movement past committee report — or continued stall — will show whether U.S. national literacy remains optional spending authority. Until then, staff the literacy product you actually need, and stop pretending a Title IV eligibility change equals a 2030 universal mandate or a code you can’t trust.
Sources
- Ministry of Education of the People’s Republic of China, National Development and Reform Commission, Ministry of Industry and Information Technology, Ministry of Science and Technology, & National Data Administration. (2026, April 2). “人工智能+教育”行动计划 [“Artificial Intelligence + Education” Action Plan]. Chinese Government. https://www.gov.cn/zhengce/zhengceku/202604/content_7065138.htm
- U.S. Congress. (2026). H.R.8747 — K–12 AI Literacy and Readiness Act of 2026 (Introduced in House text). Congress.gov. https://www.congress.gov/119/bills/hr8747/BILLS-119hr8747ih.htm
- U.S. House Committee on Education and the Workforce. (2026, July 21). Markup Action Summary. Congress.gov. https://www.congress.gov/119/meeting/house/119491/documents/HMKP-119-ED00-20260721-SD007.pdf
- Maryland General Assembly. (2026). Chapter 634 (Senate Bill 720) — Artificial Intelligence Ready Schools Act. Maryland General Assembly. https://mgaleg.maryland.gov/2026RS/Chapters_noln/CH_634_sb0720e.pdf
- University of Texas at San Antonio. (2026, September 23). Keeping the code safe: How UT San Antonio students are learning to build a safer AI. UT San Antonio Today. https://news.utsa.edu/2026/09/keeping-the-code-safe-how-ut-san-antonio-students-are-learning-to-build-a-safer-ai/
Frequently Asked Questions
What does China’s AI + Education Action Plan require by 2030?
By 2030 China aims to form deep AI–education integration, a full-stage AI education system linked horizontally across society, and a long-term nationwide AI literacy mechanism. The April 2, 2026 plan pushes K–12 AI into local curricula, AI as a higher-ed foundation course, vocational upgrades, society-wide literacy resources, teacher AI literacy standards, and security barriers for educational AI use.
What deadlines does Maryland’s Artificial Intelligence Ready Schools Act set?
The Act took effect June 1, 2026. Local school systems must adopt an AI policy within 120 days of state guidance, designate a noninstructional AI coordinator, and use a state rubric for AI tools. By June 1, 2027, AI literacy must be part of workforce and computer science standards for K–12. Teacher PD has a train-the-trainer intent by July 1, 2027.
What is OpenSecCoder?
OpenSecCoder is a free NSF-funded training platform led by UT San Antonio’s Nishant Vishwamitra. It uses sandboxed labs so students can find and fix insecure AI-generated code, including agentic coding risks. The project targets about 600 undergrads and 400 graduate students a year, with MIT-licensed materials and labs open to non-programmers.
Is H.R.8747 federal AI literacy law?
No. H.R.8747 amends ESEA so Title IV funds may support safe, effective, and responsible AI instruction and teacher professional development. The House Education and Workforce Committee ordered it reported amended 18–15 on July 21, 2026. It has not become law and does not mandate a national curriculum or local policy clocks.
How do these three literacy approaches differ for operators?
China builds universal literacy capacity with a 2030 national target. Maryland binds schools to policy, coordinator, and standards deadlines. OpenSecCoder trains critical distrust of AI-written code. Operators should map programs to one outcome — population fluency, institutional policy, or artifact verification — instead of one vague literacy label.
What should Responsible AI teams change this week?
Assign every AI training or school-AI program a named literacy outcome, owner, deadline, and failure test. For engineering teams, require a critical-evaluation gate before AI-generated code reaches production. For education operators in timed jurisdictions, staff the policy and coordinator roles before the clock expires.
#AIgenerated
Leave a Reply