AI Disciplinary Decisions: California Bans ‘Robo Bosses’
California has become the first US state to bar employers from relying solely on artificial intelligence when disciplining or dismissing workers. The development gives fresh weight to the debate over AI disciplinary decisions and the extent to which human judgment must remain at the centre of workplace justice.
In Brief
California has introduced new restrictions on AI disciplinary decisions, prohibiting employers from relying solely on automated systems to discipline or dismiss workers from July 2027 and requiring meaningful human review where AI provides the primary basis for a decision. The development provides a useful comparison for UK employers, where significant solely automated decisions are permitted in wider circumstances but remain subject to statutory safeguards and employment-law obligations. Separately, UK technology founders and investors are increasing pressure for reform of non-compete clauses and other contractual restrictions on labour mobility.
Key Points
- From 1 July 2027, California employers will be barred from relying solely on an automated decision-making system to discipline or dismiss an employee.
- Where an automated system provides the primary basis for the decision, a human reviewer must independently corroborate the outcome using relevant supporting information, and affected workers must be informed of the system’s role and the human review.
- In the UK, the Data (Use and Access) Act 2025 replaced Article 22 UK GDPR with Articles 22A to 22D, permitting significant solely automated decisions in wider circumstances but requiring safeguards including information, representations, human intervention and the ability to contest the decision. Decisions involving special category data, such as health information, remain subject to tighter restrictions.
- UK employers using AI disciplinary decisions must also comply with unfair dismissal principles, discrimination law and fair disciplinary procedures. AI should inform rather than replace accountable human judgment.
- Separately, the “Free to Start and Scale” campaign argues that non-compete clauses, lengthy notice periods and garden leave restrict skilled-worker mobility and make recruitment harder for UK start-ups and scale-ups.
- The government has not yet announced its preferred approach to non-compete reform. Existing clauses remain subject to the common-law requirement that they go no further than reasonably necessary to protect a legitimate business interest.
Governor Gavin Newsom signed the No Robo Bosses Act 2026, or SB 947, on 30 September. From 1 July 2027, California employers will be barred from relying solely on an automated decision-making system to discipline or terminate an employee. Where an automated system is the primary basis for the decision, a human reviewer must independently corroborate the outcome using relevant supporting information. For employers using AI disciplinary decisions as part of their processes, that means human involvement must be substantive rather than nominal.
The law also requires workers to be informed when an automated system has played the primary role in a disciplinary or termination decision. The employee must be told that human review took place and be given a human contact for questions. The legislation therefore goes beyond merely placing a manager at the end of an automated process: it seeks to ensure that AI disciplinary decisions involve genuine human scrutiny rather than a rubber stamp.
That is significant because Newsom vetoed an earlier version of the proposal in 2025. The revised legislation narrowed the notification requirements and the circumstances in which automated systems are restricted, while retaining the central principle that a person should not lose their job because an algorithm decided it alone. The result is a more targeted framework for AI disciplinary decisions rather than a general prohibition on workplace AI.
We considered precisely this issue in an article back in February 2026. At that stage, the central question was where employers should draw the line between AI as an analytical tool and AI as the effective decision-maker. California has now drawn that line in legislation.
What Does This Mean for UK Employers?
California's rules on AI disciplinary decisions provide a useful indication of the direction in which workplace AI regulation may develop.
UK data protection law changed in February 2026. The Data (Use and Access) Act 2025 replaced Article 22 UK GDPR with new Articles 22A to 22D, permitting significant decisions based solely on automated processing in a wider range of circumstances, provided appropriate safeguards are in place. Those safeguards include informing the individual, allowing them to make representations, obtain human intervention and contest the decision. A decision is treated as solely automated where there is no meaningful human involvement, and decisions involving special category data remain subject to tighter restrictions.
In the UK, AI disciplinary decisions must also be considered against employment-law duties that sit alongside data protection. Employers must still consider unfair dismissal principles, discrimination law and the requirements of a fair disciplinary process. A manager who simply accepts an algorithmic recommendation without independently considering the evidence, mitigation and proportionality may struggle to demonstrate that meaningful human judgment was exercised.
California’s approach therefore reinforces a practical distinction. AI can assist with evidence review, consistency and identifying relevant material, but AI disciplinary decisions become much harder to defend when the technology effectively determines the outcome. For employers, AI disciplinary decisions should be designed around accountable human judgment from the outset.
For UK employers using AI disciplinary decisions, the safest model remains one in which AI informs rather than replaces accountable human decision-making.
Non-Compete Clauses: Tech Founders Increase Pressure for Reform
Separately, the debate over non-compete clauses has intensified as technology founders and investors press the government to remove contractual barriers to labour mobility.
An open letter published under the banner “Free to Start and Scale” argues that non-competes, lengthy notice periods and garden leave are restricting the movement of skilled workers and making it harder for UK start-ups and scale-ups to recruit quickly. Signatories include founders and executives from a number of prominent technology and AI businesses.
The intervention comes while the Department for Business and Trade considers responses to its working paper on reforming non-compete clauses. Options under consideration include statutory limits on duration, a complete ban, salary thresholds and combinations of those approaches.
The new campaign also highlights a problem we identified in an article back in February 2026: reforming non-competes in isolation may simply encourage employers to rely more heavily on notice periods, garden leave or other restrictive covenants. The open letter therefore calls for wider restrictions rather than a ban limited to non-compete clauses.
There is still no announced government decision. For employers, existing non-compete clauses remain subject to the current common-law test of whether they go no further than reasonably necessary to protect a legitimate business interest. But political and commercial pressure for reform is plainly continuing to build.
Employers: What This Means
UK employers using AI in disciplinary or dismissal processes should ensure that technology supports rather than replaces accountable human judgment. California’s new legislation does not apply in the UK, but it illustrates the growing regulatory emphasis on meaningful human oversight of AI disciplinary decisions. Separately, employers using non-compete clauses should continue to review whether restrictions remain necessary and proportionate while the government considers possible reform.
- Ensure managers independently assess the evidence, mitigation and proportionality before reaching disciplinary or dismissal decisions rather than simply approving an AI-generated recommendation.
- Review automated decision-making processes against UK data protection requirements, including the safeguards applying to significant solely automated decisions and the tighter restrictions where special category data, such as health information, is involved.
- Maintain clear records showing how human judgment influenced the outcome. Employers should also consider unfair dismissal principles, discrimination risks and the requirements of a fair disciplinary process alongside data protection obligations.
- Separately, review non-compete clauses and other restrictive covenants to ensure they go no further than reasonably necessary to protect legitimate business interests, and consider how notice periods and garden leave interact with them, particularly while government reform remains under consideration.
