Intermediate level8 min readTo learn at another level, choose it before you start the course.
After this lesson you canExplain what AI rules ask employers to do at each stage of a job, from hiring to layoffs.
The short answerMany places now have rules for AI at work. Employers often must tell people when AI judges them, check AI tools for unfair results, and keep a human involved in big decisions such as firing.
In simple words
- You often have a right to know when AI helps judge you.
- AI tools must not treat people unfairly.
- Some places ban AI that reads workers’ feelings.
- Big decisions, such as firing, need a human.
The short answerWorkplace AI rules follow the job cycle: notices and bias audits in hiring, limits on monitoring, human checks on discipline and firing, and disclosure in layoffs. Where you work decides which apply, on top of general anti-discrimination law.
In simple words
- Hiring: notices, consent for video analysis, yearly bias audits in New York City.
- Monitoring: EU ban on emotion recognition at work; works councils in Germany and France.
- Decisions: limits on fully automated decisions; human checks in California from July 2027.
- Layoffs: Connecticut asks whether AI caused job cuts, and California from 2027.
The short answerEmployment AI sits under layered regimes: AI Act prohibitions and high-risk duties, GDPR Article 22, national co-determination law, and US state and city statutes on notice, bias audits and human corroboration. Compliance starts by mapping each AI tool to the decision it influences.
Key points
- Map tools to decisions: screening, assessment, allocation, monitoring, discipline, termination.
- EU: the Article 5(1)(f) ban now; Annex III point 4 deployer duties from December 2, 2027.
- US: NYC bias audits, Illinois notice and anti-proxy rule, Colorado notices and human review from 2027.
- Co-determination: German works council consent and expert rights; French CSE consultation.
Rules that apply in European Union: EU · 10 rules
Hiring 6
- Do not let AI alone make big decisions about peopleEU In force
- Help your staff understand the AI they useEU In force
- Do not use biometrics to guess race, religion or union tiesEU In force
- Explain decisions made with high-risk AI when askedEU In force
- Platforms: keep humans in charge of algorithmic managementEU From Dec 2026
- AI for hiring and managing staff becomes “high-risk”EU From Dec 2027
Watching work 6
- Do not use AI to read emotions at work or schoolEU In force
- Help your staff understand the AI they useEU In force
- Do not use biometrics to guess race, religion or union tiesEU In force
- Tell people when you use emotion or biometric AI on themEU In force
- Platforms: keep humans in charge of algorithmic managementEU From Dec 2026
- Treat face recognition and emotion AI as high-riskEU From Dec 2027
Managing and firing 6
- Do not let AI alone make big decisions about peopleEU In force
- Help your staff understand the AI they useEU In force
- Do not use AI social scores to treat people unfairlyEU In force
- Explain decisions made with high-risk AI when askedEU In force
- Platforms: keep humans in charge of algorithmic managementEU From Dec 2026
- AI for hiring and managing staff becomes “high-risk”EU From Dec 2027
Layoffs 3
- Do not let AI alone make big decisions about peopleEU In force
- Help your staff understand the AI they useEU In force
- AI for hiring and managing staff becomes “high-risk”EU From Dec 2027
Words to know
- Bias audit
- A check of whether a tool gives unfair results for some groups of people.
- Works council
- A group elected by the staff to speak for them with the employer.
- Mass layoff
- Cutting many jobs at once. Employers must often give an official notice first.
When AI helps to hire
Many companies use AI to sort job applications or score video interviews. Several places say you must be told about it.
In New York City, employers must have AI hiring tools checked for bias every year and tell candidates before use. In Illinois, employers need your consent, a clear yes, before AI analyses a video interview.
When AI watches you work
Some AI tools track how fast people work or try to read their mood. In the EU, AI that reads emotions at work has been banned since February 2025, except for medical or safety reasons.
California bans AI tools that read workers’ emotions from January 2027. In Germany and France, employers that have a works council must involve it before using AI that monitors staff.
Firing and layoffs
From July 2027, California employers may not fire or discipline workers by software alone. A human must check the result.
Some states also want to know if AI caused job cuts. Since October 2026, Connecticut asks employers whether mass layoffs are related to AI.
Try it yourself
Think of an AI tool used at work, such as one that sorts job applications. Pick your place in the diagram and read the rules for that stage.
Check yourself
A shop chain in New York City wants an AI tool to sort job applications. What must it do?
A call centre in Athens wants AI that reads workers’ moods from their voices, to judge their work. Is that allowed?
In 2028, a California company’s software picks a worker to be fired. What must happen before the firing?
Words to know
- Bias audit
- A check of whether a tool gives unfair results for some groups of people.
- Works council
- A group elected by the staff to speak for them with the employer.
- Mass layoff
- Cutting many jobs at once. Employers must often give an official notice first.
Hiring
New York City’s Local Law 144 requires an independent bias audit of automated hiring tools every year. Employers must publish a summary and tell candidates at least 10 business days before use.
Illinois requires consent before AI analyses recorded video interviews. Since 2026, it has also banned AI use that leads to discrimination in job decisions. Connecticut says an AI tool is no excuse for discrimination.
Monitoring
The EU bans AI that infers emotions at work, except for medical or safety reasons. From 2027, California bans AI tools that read workers’ emotions or collect neural data, meaning measurements from the brain and nerves.
In Germany, the works council must agree before an employer introduces systems that monitor staff behaviour or performance. In France, employers must inform and consult the works council before deploying monitoring tools.
Decisions about people
Under the GDPR, the EU’s data protection law, people have the right not to face big decisions made only by software, such as an automatic rejection. There are a few exceptions.
California’s No Robo Bosses Act, from July 1, 2027, bans firing or disciplining by software alone. California’s privacy rules add notices and opt-outs for automated decisions from 2027.
EU high-risk duties are coming
From December 2, 2027, AI for recruiting and managing staff becomes high-risk under the AI Act. Employers must then use trained human oversight, keep logs for at least six months and inform workers first. AI tools already on the market are covered only after a big change to their design.
Makers of such tools must check them against the Act’s rules themselves, add the CE mark and register them before sale. Until then, the GDPR, labour law and national works council rules already apply.
Try it yourself
Think of an AI tool used at work, such as one that sorts job applications. Pick your place in the diagram and read the rules for that stage.
Check yourself
A firm in New York City plans to use an AI tool to rank job candidates. What does Local Law 144 require?
A warehouse firm in Hamburg has a works council and wants AI that tracks how fast each worker packs. What must it do first?
A retailer in Warsaw lets software reject job applications automatically, with no person involved. What does the GDPR say?
The EU stack
Article 5(1)(f) bans emotion inference at work. From December 2, 2027, Annex III point 4 makes tools for recruitment, promotion, termination, monitoring and task allocation based on behaviour or traits high-risk, with Article 26 deployer duties.
Article 86 already gives affected workers a right to an explanation of decisions based on such AI. For gig platforms, the Platform Work Directive adds human control of algorithmic management; EU countries must put it into national law by December 2, 2026.
National labour law
Germany’s Works Constitution Act requires works council consent for monitoring systems and for AI-assisted selection guidelines, and lets the council call in an AI expert.
France requires the works council (CSE) to be informed and consulted before new technologies and monitoring tools. Candidates must be told which assessment methods, including AI, are used.
US state and city law
NYC Local Law 144 requires a bias audit within the past year and 10 business days’ notice. Illinois bans discriminatory AI use, ZIP codes as proxies for protected groups and undisclosed AI use in job decisions.
California requires automated-decision data to be kept for four years and bans software-only discipline or firing from July 2027. Colorado’s SB 26-189 requires notices and human review from 2027.
Layoffs and disclosure
California’s SB 951 requires mass-layoff notices to flag a technology displacement from 2027, naming the jobs, the tasks automated and the technology used. Connecticut asks with each mass-layoff notice whether AI played a part.
Both rules require disclosure only; neither restricts the layoffs themselves.
Try it yourself
Think of an AI tool used at work, such as one that sorts job applications. Pick your place in the diagram and read the rules for that stage.
Check yourself
In 2028, a warehouse operator in Antwerp starts using a newly released AI tool that allocates tasks based on each worker’s past performance. How does the AI Act classify this?
An employer hiring in Chicago lets its AI screening model weigh applicants’ ZIP codes. What is the problem under Illinois law?
In November 2026, a Connecticut employer files a federal mass-layoff notice, and AI tools replaced some of the jobs. What must it do?
Sources
- Automated employment decision tools (NYC Department of Consumer and Worker Protection)
- SB 947, No Robo Bosses Act (California Legislature)
- AI Act, Article 5: prohibited practices (EU AI Act Service Desk)
- GDPR, Regulation (EU) 2016/679 (EUR-Lex)
This lesson was generated by AI systems under the editorial rules of Silicon AI News and checked against the sources it lists. The live parts come from our checked stories, trackers, model comparison and rules checker.