The jobs AI law already changed before any AI arrived

The EU AI Act assigns human oversight to a natural person

Since 2 August 2026 the EU AI Act requires deployers of high-risk AI systems to assign human oversight to natural persons with the competence, training and authority to do it. Not to a committee, not to a policy — to people, who must be given the authority. Where the deployer controls the input data, it must also ensure that data is relevant and sufficiently representative.

Two of the Act's high-risk categories name work directly. Recruitment and selection of people is one, which is why it sits on the recruiter's page. Since August 2026, systems used to allocate tasks based on individual behaviour or personal traits are treated as high-risk too — that is the first-line manager's job described in a statute.

The Act also splits the duty in two, and the split lands on two different jobs. Article 26 is the deployer's half — the employer running the system. Article 50 is the provider's half: from the same date, a system meant to interact with people must tell them they are dealing with an AI, and synthetic audio, image, video or text must be marked in a machine-readable format that is effective, interoperable and robust. The second of those is a build requirement with an engineering cost, and the person who has to write it into a spec is a product manager.

The practical consequence is not that these roles get harder to automate. It is that a duty now exists which did not before, it is legally attached to a person, and somebody has to hold it. That is a change to the job, produced entirely by a rule.

A right to a human being on the other side

Colorado's SB26-189 repealed and re-enacted the state's 2024 AI Act, giving a person the right to meaningful human review after an automated decision. Korea's AI Framework Act, in force in its amended form from 21 July 2026, classifies AI used in certain public and medical settings as high-impact and attaches obligations to it. Illinois barred AI from delivering therapy services. California barred AI systems from using titles such as nurse or doctor.

Note what the last two regulate: not the technology's accuracy, but who is allowed to say they are doing the work. That is a boundary drawn around an occupation rather than around a model, and it is the kind of rule that does not appear in any capability benchmark.

Record the verb strength, because it differs and the difference decides everything: Korea's Article 34 duty and its Article 35 best-efforts language are not the same instrument, and a Chinese rule saying 原则上 is not a flat requirement. A headline that says 'country X now requires' is frequently reading a best-efforts clause.

And one guardrail many people assume exists, which does not

California's SB 7, the 'No Robo Bosses Act', would have barred employers from relying solely on an automated system to discipline or dismiss a worker. The legislature passed it; the Governor vetoed it on 13 October 2025; the veto was sustained on 2 March 2026. As of that date California has no such rule.

This gets its own stage on this site — 'rule not enacted' — rather than being filed under constraint, because constraint means regulation suppressing adoption and a defeated bill is the failure of one. It establishes something real and checkable about the legal environment, and nothing at all about what employers are doing.

Regulation that requires adoption, not just restrains it

Almost all Western coverage of AI regulation treats it as a brake. In the evidence base it is also an accelerator, and disproportionately so in the market our Chinese readers are in. China's education ministry issued a guideline requiring AI general education in schools and AI teaching ability in teacher training; a five-ministry action plan then required AI to be written into teacher qualifications. The US Office of Management and Budget's Memorandum M-25-21 requires every executive department and agency to act. A 2022 State Council requirement pushed cross-provincial government services onto a national platform.

This is why the stage machine has a 'mandate' category at all. Until it was added, the model could push adoption down but not up — which systematically excluded policy-driven markets. A mandate is weighted cautiously and never on its own: it shows adoption is being required, not that it has happened.

And a state can create work simply by naming it. China's labour ministry put 17 new occupations and 42 new work-types out for comment, then 12 further ones including an embodied-AI robot application technician. A classification is a different and often better kind of evidence than a company's rollout: it is the state saying this work exists, at a scale worth counting.

Courts, unions and registers write rules too

Florida's Supreme Court amended its rules so that signing a filing is itself a representation about the legal authority cited in it. IATSE's 2024 Basic Agreement added Article XLIX, keeping work performed by prompting or overseeing a system inside the bargaining unit — a union contract deciding that operating the tool is still the job. China's transport ministry requires an onboard safety officer in conditionally and highly automated vehicles, and says autonomous road freight should in principle carry one.

Registers are evidence as well. China's cyberspace regulator reported 868 filed generative AI services and 530 registered, with a duty to display the filing number on any live application. Ontario expanded pharmacists' prescribing scope while dispensing automates — the scope of an occupation widening by regulation at the same time as part of it narrows by machine.

Everything cited here

Each one opens the full record: its source and source tier, the dates, the scope it applies to, who verified it and what it does not establish.

1. Since 2 August 2026 the EU AI Act requires deployers of high-risk AI systems to assign human oversight to natural persons with the necessary competence, training and authority (Article 26) — Regulation (EU) 2024/1689 (Artificial Intelligence Act), Article 26 — verbatim mirror of the Official Journal text
2. Since 2 August 2026 the EU AI Act requires that, where a deployer controls the input data of a high-risk system, it must ensure that data is relevant and sufficiently representative — Regulation (EU) 2024/1689 (Artificial Intelligence Act), Article 26(1) and 26(4) — verbatim mirror of the Official Journal text
3. The EU AI Act (in force 1 August 2024) lists AI used to recruit or select people — placing targeted job ads, filtering applications, evaluating candidates — as high-risk under Annex III, point 4 — Regulation (EU) 2024/1689 (Artificial Intelligence Act), Annex III point 4 — verbatim mirror of the Official Journal text
4. Since 2 August 2026 the EU AI Act treats AI systems used to allocate tasks based on individual behaviour and to monitor and evaluate worker performance as high-risk under Annex III point 4(b) — Regulation (EU) 2024/1689 (Artificial Intelligence Act), Annex III point 4(b) — verbatim mirror of the Official Journal text
5. From 2 August 2026 the EU AI Act requires that a system meant to talk to people tells them it is an AI, and that synthetic output is machine-readably marked — EU AI Act, Article 50 (verbatim mirror of Regulation (EU) 2024/1689)
6. Colorado's SB26-189 gives a person the right to meaningful human review after an automated consequential decision goes against them, with a plain-language explanation due within 30 days — Colorado General Assembly, SB26-189 Automated Decision-Making Technology — session law chapter 131, signed 2026-05-14
7. Colorado enacts SB26-189, repealing and re-enacting its 2024 AI Act: after an automated system makes a consequential decision with an adverse outcome, the person may demand meaningful human review — Colorado General Assembly, SB26-189 Automated Decision-Making Technology — session law chapter 131, signed 2026-05-14
8. Korea's AI Framework Act, in force in its amended form from 21 July 2026, classifies AI used for public-service eligibility decisions as high-impact and requires human management and supervision of it — 국가법령정보센터 (Korea Ministry of Government Legislation) — 인공지능 발전과 신뢰 기반 조성 등에 관한 기본법, 법률 제21311호
9. Illinois enacted the Wellness and Oversight for Psychological Resources Act, barring AI from mental health and therapeutic decision-making and permitting it only for administrative support — Illinois Department of Financial and Professional Regulation (state agency announcement)
10. California enacted AB 489, barring AI systems from using titles such as 'nurse' or 'doctor' where this implies a licensed person is providing the care — California Legislative Information (AB 489, chaptered)
11. California's SB 7 would have barred relying solely on an automated system to discipline or dismiss a worker; it passed the legislature, was vetoed, and the veto was sustained on 2 March 2026 — California State Legislature — SB 7 (2025–2026), bill status and Governor's veto message (leginfo.legislature.ca.gov)
12. China's education ministry advisory committee issued an AI general-education guideline and a generative-AI use guideline for all primary and secondary schools — 中华人民共和国教育部 科技与信息化司——《2025 年 5 月教育信息化和网络安全工作月报》
13. Five Chinese ministries issue an AI-plus-Education action plan requiring AI to be written into the teacher qualification exam and certification, with tiered training to reach every teacher — 教育部等五部门《关于印发「人工智能+教育」行动计划的通知》教科信〔2026〕1号(教育部政府门户网站)
14. US OMB Memorandum M-25-21 requires every executive department and agency, including independent regulators, to identify a Chief AI Officer and allocate resources and responsibilities to the role — US Office of Management and Budget, Memorandum M-25-21 (3 April 2025), signed by OMB Director Russell T. Vought
15. China's State Council General Office required cross-provincial government services to run through the national integrated platform, with listed items standardised nationwide by mid-2023 — 国务院办公厅 · 国办发〔2022〕34号(中国政府网)
16. China's labour ministry put 17 new occupations and 42 new work-types out for comment, adding a 'generative AI system tester' work-type under the existing 'generative AI system operator' occupation — 中国就业网(人力资源和社会保障部)——《职业上新!42 个新工种亮相》,2025-05-09
17. China's labour ministry put 12 further new occupations out for comment, among them 'embodied-AI robot application technician' — 中国就业网(人力资源和社会保障部)——《12 个新职业向社会公示》,2026-07-10
18. Florida's Supreme Court amends Rule 2.515(d)(2): signing a filing represents that the legal authorities cited exist and are accurate, with sanctions available, effective June 15, 2026 — Supreme Court of Florida, No. SC2026-0673 — In re: Amendments to Florida Rule of General Practice and Judicial Administration 2.515 (2026-05-28)
19. IATSE's 2024 Basic Agreement added Article XLIX, keeping work done by prompting or overseeing an AI system inside covered union work — IATSE — 2024 Basic Agreement MOA, Article XLIX (fully executed; copy hosted by IATSE Local 728)
20. China's transport ministry says autonomous road freight should in principle carry an onboard safety officer, and bans autonomous vehicles from carrying dangerous goods — 中华人民共和国交通运输部办公厅 —《自动驾驶汽车运输安全服务指南(试行)》,交办运〔2023〕66 号
21. China's transport ministry requires an onboard safety officer in conditionally and highly automated taxis, and caps remote safety officers at one per three fully driverless taxis — 中华人民共和国交通运输部办公厅 —《自动驾驶汽车运输安全服务指南(试行)》,交办运〔2023〕66 号
22. China's cyberspace regulator reported 868 filed generative AI services and 530 registered applications as of 30 April 2026, and requires every live application to display which filed service it uses — 国家互联网信息办公室——《关于发布生成式人工智能服务已备案信息的公告(2026 年 3 月至 4 月)》
23. Ontario expanded pharmacists' scope from July 2026 to administer six more publicly funded vaccines and to assess and prescribe for nine more common ailments, taking the total to 33 — Government of Ontario