South Korea's government has previewed a draft AI Legislation Framework built around the AI Basic Act
The draft is a set of four directions for future AI legislation, reported to the 5th Chief AI Officers Council on 2 October 2026. It is not law and the release gives no text or timetable, so it is a signal to track, not a duty to meet.
Bottom line: The National AI Strategy Committee reported a draft framework on the government-wide direction of AI legislation after the AI Basic Act took effect. A draft reported to a council is not binding on anyone.
Who this affects: Compliance, policy and public-affairs teams at AI providers and deployers active in Korea.
Issue date: 2 October 2026 (release date). No effective date is stated.
What changed: The government disclosed four basic directions and legislative principles for AI development, convergence, safety and trust, and misuse prevention. It did not publish the framework text in the release.
Analysis: In our reading, the useful signal is the order of priorities. Business predictability sits beside harm response and security-sector gaps, which suggests follow-on bills could address more than the AI Basic Act alone. The release does not confirm that.
- Instrument (English)
- Draft "Republic of Korea AI Legislation Framework" (agenda item 1, 5th CAIO Council)
- Authority
- National AI Strategy Committee, chaired by the President; released through the Personal Information Protection Commission (PIPC) as a joint-ministry press release
- Jurisdiction
- South Korea
- Current Status
- Draft (안) reported to the council for discussion
- Bindingness
- Not applicable (non-binding). A draft reported to an inter-agency council; the release states no legal effect
- Issue date
- 2 October 2026 (press release and council date)
- Effective date
- Not stated in the source
- What changed
- The committee reported a draft with four basic directions and field-by-field legislative principles on AI development, AI convergence, AI safety and trust, and prevention of AI misuse
- Who this affects
- Agencies preparing AI-related legislation, and AI providers, deployers and their advisers who track Korean policy. The release names no regulated parties
- Scope
- Government-wide direction of AI legislation after the AI Basic Act took effect. The release also covers three other agenda items: the Everyone's AI service, a public data pipeline and public-sector privacy management
- Primary source
- PIPC press release, 5th CAIO Council (Korean), 2 October 2026
- Workflow impact
- Not stated in the source. Nothing in the release creates a filing, disclosure or approval step
- Operational next step
- Add the framework to your Korea watch list, confirm which AI Basic Act duties already apply to you, and wait for published text before changing policy
- Last verified
- 2026-10-25
Primary source: PIPC press release, 5th CAIO Council (Korean)
Key takeaways
- The framework is labelled a draft and was reported to a council on 2 October 2026. The release does not call it adopted.
- Its four directions, in the release's words:
①AI기본법 중심 입법, ②국방·안보 분야 입법 공백 해소, ③사업자 예측가능성 제고, ④역기능 발생 시 즉시 대응
- The release gives no text, bill number, consultation or timetable.
- The same council meeting took up the Everyone's AI service, a public data pipeline and public-sector privacy management.
Is the AI Legislation Framework law?
No. The release titles it 「대한민국 인공지능 입법 프레임워크(안)」, and the final character marks it as a draft. The National AI Strategy Committee, which the release says is chaired by the President, reported it to the 5th Chief AI Officers (CAIO) Council on Friday 2 October 2026 at 4 p.m. in Seoul. The release describes the draft as the government-wide direction for AI legislation, prompted by a sharp rise in demand for AI-related legislation after the AI Basic Act took effect.
The release does not say the council approved or adopted it, and it names no bill, public comment period or date. Treat it as a stated policy direction. For duties that apply today, see our separate piece on high-impact AI duties under the AI Basic Act.
What are the four basic directions?
The release lists them in one sentence: ①AI기본법 중심 입법, ②국방·안보 분야 입법 공백 해소, ③사업자 예측가능성 제고, ④역기능 발생 시 즉시 대응
In English: legislation centred on the AI Basic Act, closing legislative gaps in defence and security, raising predictability for business operators, and responding immediately when harmful effects occur.
Under those directions the draft sets out legislative principles by field. The release names four: AI development, promotion of AI convergence, securing AI safety and trust, and preventing AI misuse. It does not describe what any principle would require, which sectors are meant by the defence and security gaps, or what an immediate response mechanism would look like.
For a provider, the second and third directions point in different ways. Closing gaps suggests new rules in some areas, while predictability suggests clearer ones.
What else did the council take up on 2 October?
The meeting was the first chaired by Senior Presidential Secretary for AI Future Planning Lee Hae-min. The release says CAIOs from 42 central administrative agencies attended, together with 13 chairs of subcommittees, special committees and task forces, with substitutes included. Attendance was widened from 28 minister-level agencies to 42 by adding 14 vice-minister-level bodies, including the personnel agency, the legislation ministry and the national data agency named in the release.
Three further agenda items were discussed. The Ministry of Science and ICT (MSIT) said the Everyone's AI service is due for full public launch in December, and asked partner agencies to link open public APIs, open more APIs and data, set criteria for how AI agents process personal data, and verify service security and safety. The Ministry of the Interior and Safety asked agencies to register high-value data after the government-wide data pipeline opened on 18 September. The PIPC, citing recent large public and private hacking and leak incidents, asked for ministry-level privacy and security management systems, a stronger role for chief privacy officers, and risk-based identification of information assets.
A written progress report on the first-half 2026 AI action plan was also submitted. The release gives no figures from it.
| Date | Event in the release |
|---|---|
| 18 September 2026 | Government-wide data pipeline opened, per the Ministry of the Interior and Safety item |
| 2 October 2026, 4 p.m. | 5th CAIO Council met in Seoul; draft framework reported as agenda item 1 |
| December 2026 | Everyone's AI service due for full public launch, per MSIT |
| Not stated | Effective date, bill, consultation or adoption date for the framework |
What should compliance and policy teams do now?
Keep the draft in view without acting on it as if it were a rule. Nothing in the release creates a duty.
- Add the framework and the four directions to your Korea regulatory watch list, with the PIPC release as the source of record.
- Confirm which AI Basic Act and enforcement-decree duties already apply to your products.
- If you work on defence, security or public-sector AI, flag the stated legislative gap as an area where rules may change, and ask counsel in Korea what to monitor.
- If you supply AI agents to Korean public bodies, note the MSIT request on personal-data criteria and security verification for the Everyone's AI service.
FAQ
Is the Korean AI Legislation Framework law?
No. The government release labels it a draft and says the committee reported it to the council. The release does not describe it as adopted, and it states no effective date.
What are the four basic directions in the framework?
According to the release, they are AI Basic Act-centred legislation, closing legislative gaps in defence and security, raising business predictability, and immediate response when harms occur.
Does the framework change obligations for AI providers in Korea now?
The release does not say so. It reports a legislative direction, not new duties. Obligations under the AI Basic Act and its implementing rules are the instruments to check for current duties.
Who attended the 5th CAIO Council?
The release says Chief AI Officers from 42 central administrative agencies attended, plus 13 chairs of subcommittees, special committees and task forces, with substitutes counted. Senior Presidential Secretary Lee Hae-min chaired for the first time.
When does the Everyone's AI service launch?
The release says MSIT told the council the service is due for a full public launch in December, and asked partner agencies for cooperation on APIs, data, personal-data criteria for AI agents and security checks.
What we did not verify
What we opened: the PIPC press release (board BS074, post 12556, dated 2 October 2026), in Korean, on 25 October 2026.
What we did not open: the draft framework itself, the council agenda papers, the PDF and HWPX attachments to the release, any other ministry's version of the release, and any later statement on timing. The release names the committee and the PIPC as publishers; we have not checked which body hosts the original on its own site. English renderings of Korean text are ours.
What we refuse to claim: that the framework is adopted or binding, what any principle would require, which bills will follow, or any effect on current AI Basic Act duties.
South Korea's AI Legislation Framework is a draft direction, not a rule. Monitor it, confirm your current AI Basic Act duties, and revisit your plans when text or a bill is published.
Related briefings
Back to the AI Regulation News hub or the Asia-Pacific tracker.
Source File
Primary: https://www.pipc.go.kr/np/cop/bbs/selectBoardArticle.do?bbsId=BS074&mCode=C020010000&nttId=12556
Secondary: None relied on for the facts above.
Last checked: 2026-10-25 (UTC)
Status: Draft policy framework, not binding.
Editorial note: Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.
Informational analysis for working professionals, not legal advice. Confirm how any rule applies to your situation with qualified counsel.
Sponsored Training
Practical AI training for regulated professionals, built around verification, documentation and a defensible process. See the courses.