A bill introduced on 3 August 2026 by Rep. Kim Seong-won and nine other lawmakers would amend the AI Framework Act to give the government a legal basis to designate and manage "high-risk AI models" — those posing a serious risk to national security or public safety — and to build a cyber-threat response system for high-impact AI. The bill has been referred to the competent National Assembly committee and has not yet been enacted; no effective date or penalty is specified in the current text. It would primarily affect developers and operators of AI models that the government designates as high-risk on national-security or public-safety grounds.

What the bill would add

  • New designation power (Art. 33-2, new). The bill would let the state identify and manage specific AI models found to pose a material risk to national security or public safety, separate from the Act's existing general rules for "high-impact" AI.
  • New response-system duties (Art. 34(1)6 and 7, new). The bill would require a nationwide cyber-threat response system for high-impact AI, addressing risks such as AI-enabled cyberattacks, large-scale disinformation generation and distribution, and breaches of critical national data.
  • No threshold, effective date, or penalty is set out in the bill summary. These are the gaps to watch as the bill moves through committee — the current text does not specify who qualifies as "high-risk," when the duty would start, or what non-compliance would cost.

What this means for you

  • This is a bill, not current law. Nothing changes yet under the AI Framework Act as a result of this proposal; treat it as a signal of direction, not an obligation.
  • If your AI model could plausibly be framed as security- or safety-relevant (e.g., cybersecurity tooling, large-scale content generation, critical-infrastructure-adjacent applications), flag this bill for tracking now, since a future designation power could reach models that today sit outside the Act's general high-impact category.
  • Watch for the threshold and penalty once drafted. The bill as summarized does not yet fix who is covered or what the exposure is — check back before assuming you are in or out of scope.
  • Re-check your position at committee markup and at any second reading. Designation criteria, effective dates, and penalties are typically added or narrowed during committee review, and that is where the practical scope of this duty will actually be set.

Source: (no source URL provided in the original record.)