Kenya

Information uploaded as at September 2026

AT A GLANCE

In Kenya, AI is used in the courts and law enforcement. Following the rapid, COVID-19 pandemic-driven digitisation of the court system, the Judiciary has begun deploying automated transcription tools, an AI legal-research platform called Hakimu.ai, and a public-facing judicial chatbot. The Chief Justice of the Kenyan Supreme Court has reported that AI now handles most court transcriptions, with accuracy improving from about 30% ‘when [it] started’ to roughly 80% as of August 2025, as the system adapted to local accents, intonations, and courtroom phrasing. In law enforcement, the National Police Service operates facial-recognition technology through its urban CCTV network.

Despite this growing use of AI, Kenya has no AI-specific statute in force. However, following its initial announcement in August 2025, the Judiciary developed a Draft Judiciary Artificial Intelligence Policy, with Draft 2.0 last circulated as at June 2026. The Draft Policy requires, among other things, human oversight, disclosure of AI use, and independent verification of AI-assisted filings. Kenya has also adopted a National AI Strategy 2025–2030, and an Artificial Intelligence Bill (2026) is in progress. In the meantime, the Data Protection Act (2019), the Computer Misuse and Cybercrimes Act (2018), the Evidence Act, and the fair-trial and privacy guarantees under the Constitution of Kenya (2010) apply.

USE

Law enforcement

Operational support

Law enforcement entities reportedly use AI to assist with the recruitment process for police officers.

Data review and analysis

An AI live facial-recognition system was launched by the National Police Service (‘NPS’) on its urban CCTV network in 2018, allegedly using NeoFace technology by NEC Corporation, a Japanese telecommunications company. This system draws on feeds from thousands of cameras, especially in major cities such as Nairobi and Mombasa. The system issues alerts to officers when a suspect is identified. In June 2026, the Interior Ministry announced plans to link the CCTV network to the national identity database so that faces captured on camera can be matched against official ID photographs. The facial recognition system has also been rolled out in transport licensing and digital identity systems to establish a broad digital identity infrastructure.

Kenya’s Integrated Command and Control System incorporates licence plate recognition, traffic and road surveillance and phone metadata and mobile phone interception, such as through International Mobile Subscriber Identity (‘IMSI’) catchers or stingray-style devices, to track users’ locations.

Prosecutors

As at September 2026, there are no reported cases of prosecutors in Kenya using AI. However, the Office of the Director of Public Prosecutions operates the Uadilifu Case Management System, which offers customisable workflows and automated notifications to manage prosecution files. Uadilifu has been integrated with the Judiciary’s e-filing system, allowing prosecutors to file charge sheets, submissions and affidavits electronically and generating unique reference numbers for police files.

Courts

Case management

In 2025, the Kenyan judiciary undertook a pilot programme, in which proceedings were recorded digitally and transcribed overnight by AI. The Chief Justice of the Supreme Court said that transcription is ‘majorly done by AI’, with accuracy rising from around 30% to 80% (as at August 2025), as the model adapted to local accents, intonations, and courtroom phrasing. The Judiciary reported delivering over 154,000 legal transcripts across 126 court stations. As at September 2026, there are plans for a national transcription centre.

Legal research, analysis and drafting support

Hakimu.ai is an AI-powered legal search engine, introduced by the Kenyan judiciary in 2026. The tool aims to help judges, researchers and legal professionals access and understand Kenyan case law.

Defence

Legal research, analysis and drafting support

As at September 2026, there is no specific AI tool available to defence counsel. However, lawyers have access to legal AI tools, such as Hakimu.ai, to assist with their research. Moreover, legal tech startups including Lawfly AI are building tools specifically designed for the Kenyan legal environment, with regular updates on local legal documentation.

Victims

Victims generally only have indirect standing in Kenya to participate in criminal proceedings. Section 9(1) of the Victim Protection Act establishes the rights a victim has during the trial process, including to be present at trial in person or through a representative of their choice, to give views in plea bargaining, and to be informed in advance of the evidence the parties intend to rely on. Victims may also participate directly pursuant to Section 88 of the Criminal Procedure Code, which allows a judge to permit the prosecution to be conducted by any person.

As at September 2026, there are no reported uses of AI by victims in criminal proceedings in Kenya.

TRAINING

There is no systematic, mandatory AI-focused training for Kenyan judges, prosecutors, law enforcement or defence counsel on the responsible use of AI. There are, however, several optional training programmes available at all levels within the justice system:

REGULATION

Kenya has no binding AI-specific law for criminal proceedings. Regulation instead rests on general statutes, including evidentiary rules on computer-generated evidence, data protection restrictions on automated decision-making, and constitutional fair trial guarantees. The Computer Misuse and Cybercrimes Act 2018 was narrowed in March 2026 when the Court of Appeal struck down offences previously used against AI-generated content as unconstitutionally vague. Reform is imminent, however, with the AI Bill 2026 before the Senate proposing a risk-based regulatory framework, and a Draft Judiciary AI Policy mandating human oversight of high-consequence judicial AI tools.

AI Regulations

Kenya’s general AI strategy is set out in the National AI Strategy 2025–2030, launched in March 2025 by the Ministry of Information, Communications and the Digital Economy. The Strategy is a non-binding policy that points to future legislation, including risk classification of AI systems and stronger data governance. The Strategy sets out key action points for the development of Kenya’s AI regulatory framework, including:

1. Reviewing sectoral laws to align with AI governance;
2. Harmonising East and Central Africa’s digital, tax and cybersecurity laws for cross-border data compliance;
3. Implementing a soft AI regulatory framework to balance innovation while enhancing safeguards;
4. Developing AI and emerging technology legislation as AI adoption matures; and
5. Leveraging regulatory sandboxes to refine AI governance and compliance standards.

At the regional level, Kenya’s National AI Strategy aligns with the African Union’s AI Continental Strategy, launched in June 2024, which aims to leverage AI’s transformative potential to accelerate socio-economic growth and development across Africa.

Guidelines for practitioners

As at September 2026, there are no AI-specific guidelines for Kenyan practitioners. However, under the Law Society of Kenya’s Code of Standards of Professional Practice and Ethical Conduct (2017), advocates are bound by professional obligations of competence, diligence (SOPPEC-3), and candour to the court (Section 152), which may require advocates to check AI-assisted research and submissions for errors, and take responsibility for the documents submitted to court.

Criminal procedure rules

Evidence Act (Cap 80) (1963)

The admissibility of computer-generated and other electronic evidence is governed principally by sections 106A and 106B of the Evidence Act. Section 106A establishes that the contents of electronic records may be proved in accordance with Section 106B. Section 106B(1), in turn, treats a ‘computer output’ as a document admissible if statutory conditions are met and requires a certificate identifying the record and the device and vouching for the reliability of the producing computer. Section 106B(2) establishes the four conditions for computer output to be deemed admissible: (i) the output was produced by the computer during the period over which it was used to store or process information for activities regularly carried on over that period by a person having lawful control over its use; (ii) during that period, information was regularly fed into the computer in the ordinary course of those activities; (iii) the computer was operating properly; or, if it was not, any malfunction or downtime was not such as to affect the electronic record or the accuracy of its content; and (iv) the information in the record reproduces, or is derived from, information fed into the computer in the ordinary course of those activities.

These rules would be central to any challenge to AI-generated or AI-reviewed evidence, and to authenticating audio, video or images that may have been synthetically produced.

Data protection legislation

Data Protection Act (2019)

The Data Protection Act is Kenya’s data privacy law, enacted to give effect to the constitutional right to privacy by regulating the collection, processing, storage, and transfer of personal data. Section 35 gives every data subject the right not to be subject to a decision based solely on automated processing (including profiling) that produces legal effects or significantly affects them, subject to limited exceptions (contract, authorising law with safeguards, or consent), and requires the controller to notify the data subject where such a decision is taken. These requirements may be relevant to the development and deployment of AI tools in criminal justice.

The Act also requires a data protection impact assessment before high-risk processing, which includes large-scale processing of sensitive data, systematic monitoring of publicly accessible areas, and automated decision-making and profiling.

Cybersecurity laws

Computer Misuse and Cybercrimes Act (2018)

The Computer Misuse and Cybercrimes Act criminalises a range of digital offences and provides investigative powers for the search, seizure and interception of digital evidence.

Part III establishes several offences spanning unauthorised access, interference and interception of computer systems and data (Sections 14 to 17), trafficking in illegal devices and access codes (Section 18), computer forgery and computer fraud (Sections 25 and 26), identity theft and phishing (Sections 29 and 30), and the wrongful distribution of obscene or intimate images (Section 37), with enhanced penalties where the target is a protected computer system such as banking, payment or emergency infrastructure (Section 20).

On 6 March 2026, the Court of Appeal, in Bloggers Association of Kenya (BAKE) v Attorney General [2026] KECA 430 (KLR), declared sections 22 and 23 (the ‘false publication’ and ‘false information’ offences) unconstitutional as overly broad and vague, unjustifiably limiting freedom of expression and media freedom, while upholding the Act’s court-ordered surveillance and data-access powers subject to specificity and proportionality. These provisions had previously been used to prosecute the sharing of AI-generated images and other allegedly false online content (see ‘Cases’ section below).

Human rights

Constitution of Kenya (2010)

Human rights law in Kenya is underpinned by the Constitution. The following articles may be relevant in the context of AI in criminal proceedings:

• The right to a fair hearing and fair trial (Article 50).
• The right to privacy (Article 31).
• Equality and freedom from discrimination (Article 27).
• The right to fair administrative action, including written reasons for decisions that adversely affect a person (Article 47).

The Constitution specifically guarantees an accused person the right to be informed in advance of the prosecution’s evidence and to adduce and challenge evidence (Article 50(2)(j)–(k)), which supports disclosure of, and the ability to test, any AI-assisted material relied on by the prosecution.

Kenya is also a party to the International Covenant on Civil and Political Rights (‘ICCPR’). Relevant rights include: the right to a fair trial, to judicial independence, and to express one’s views and have them considered in legal proceedings (Article 14 ICCPR); the right to non-discrimination (Articles 4 and 14 ICCPR); and the right to privacy and information self-determination and data protection (Article 17 ICCPR).

Kenya has also ratified the Convention on the Rights of the Child, which contains relevant provisions, including:

• Article 40 – Juvenile justice: requiring that a child alleged, accused of, or recognised as having infringed the penal law be treated in a manner consistent with the promotion of the child’s sense of dignity and worth, reinforcing respect for the human rights of others, and taking into account the child’s age and the desirability of reintegration.
• Article 16 – Right to privacy: providing that no child shall be subjected to arbitrary or unlawful interference with their privacy, family, home or correspondence, nor to unlawful attacks on their honour and reputation, and that the child has the right to the protection of the law against such interference.

Outlook

Artificial Intelligence Bill (2026)

As at September 2026, the principal AI legislative initiative in Kenya is the Artificial Intelligence Bill (Senate Bill No. 4 of 2025), which received its first reading in the Senate on 2 April 2026. Modelled in part on the EU AI Act, the Bill would:

• Establish an independent Office of the AI Commissioner under Article 260(q) of the Constitution, headed by a Commissioner appointed via a Public Service Commission–led process (nominated by the President and confirmed by Parliament) for a five‑year renewable term (Part II). The Commissioner’s functions include implementation and enforcement, risk assessment, conformity audits, post‑market surveillance, standard‑setting, management of regulatory sandboxes, complaint investigation, and maintenance of a public register of high‑risk AI systems, with powers to inspect premises, demand information, issue enforcement notices, impose administrative fines, and summon witnesses (Part II).
• Establish a multi‑stakeholder Advisory Committee on AI, chaired by the Commissioner and including representatives from the Ministry of ICT, the Office of the Data Protection Commissioner, National Commission for Science, Technology and Innovation, civil society, gender‑balanced county nominees, and independent ethics or human‑rights experts. The Committee would advise on emerging risks and review proposed guidelines and regulations (Part III).
• Adopt a four‑tier, EU‑style risk framework (unacceptable, high, limited, minimal), requiring the Commissioner to classify AI systems by their risk to health, safety, fundamental rights, the environment, or societal welfare (Part V).
• Provide that detailed classification criteria, assessment procedures, and sandbox conditions are left to subsidiary regulations made by the Cabinet Secretary in consultation with the Commissioner — a design choice that has drawn commentary for creating uncertainty about where a given system will fall until those regulations are issued (Parts V and VI).
• Provide that high‑risk sectors include healthcare, education, agriculture, finance, security, employment, and public administration (Part V). Providers and deployers of high‑risk systems must carry out pre‑deployment risk and human‑rights impact assessments with mitigation measures, ensure transparency and explainability of decision‑making, maintain documentation and datasets, comply with the Data Protection Act, implement cybersecurity safeguards, and obtain consent for use of a person’s likeness (Part V).
• Provide that non‑compliance attracts criminal penalties: fines of up to five million shillings or up to two years’ imprisonment for the most serious offences (for example, deploying unacceptable‑risk systems or high‑risk systems without required assessments), with lesser fines or six‑month terms for transparency or obstruction offences, and liability extending to directors who knowingly fail to ensure compliance (Part VI).
• Give individuals a right to an explanation and a right to human review of significant automated decisions (clauses 26(1)(c), 28).
• Require labelling of AI‑generated content and criminalise harmful deepfakes intended to deceive, defame or incite violence (clauses 26, 35–36).

Draft Judiciary AI Policy (2026)

The Draft Judiciary AI Policy, published in June 2026, envisages using AI to automate case triage and hearing scheduling so that judicial officers can focus on substantive matters. This proposed framework is designed to guide the Judiciary’s responsible adoption of AI as part of its broader digital transformation agenda. It establishes principles such as transparency, explainability, human oversight, fairness, and data protection, creates a risk-based governance framework for AI systems, and emphasises that AI may support, but never replace, judicial decision-making. Under the framework, high-consequence tools (i.e., predictive sentencing) require mandatory human oversight and audits; medium-risk tools (i.e., legal-research tools) require verification; and low-risk tools (i.e., communication bots and workflow automation) face general monitoring.

The plan includes ICT help desks at Huduma Centres and courts to assist vulnerable and digitally inexperienced users, addresses interoperability by linking systems with justice-sector agencies, and strengthens cybersecurity. Announced during a Kenya–Ethiopia judicial exchange, the initiative also positions Kenya to share its ICT expertise with Ethiopia, which aims for full judicial digitisation by 2026.

As a draft policy, it is not binding as at September 2026 and contemplates phased implementation through governance structures, pilot projects, and judicial training.

Kenya Bureau of Standards, Draft AI Code of Practice (2024)

The Kenya Bureau of Standards published a Draft Information Technology AI Code of Practice in 2024 to help organisations develop and use AI responsibly. The Code emphasises transparency, explainability and controllability to protect citizen rights, and identifies common engineering pitfalls and threats. As at September 2026, the Code remains in draft form.

CASES

AI misuse in court filings

In Nayan Mansukhlal Savla v Commission on Administrative Justice & Kenya Psychiatric Association (HCJRMISC/E120/2025), the High Court annulled a judgment delivered in December 2025 after it was revealed that the foundational documents (i.e. the pleadings) were the product of AI tools. The Court noted that the drafting of pleadings is strictly regulated by the Civil Procedure Rules, and that since these rules do not authorise AI-generated content, such documents are considered procedurally deficient. Moreover, Justice Chigiti held that using AI may provide an ‘unfair advantage’ to one party, which undermines the adversarial nature of the Kenyan legal system. The ruling also described the use of AI as an affront to access to justice, as guaranteed under Article 48 of the Constitution.

Deepfakes

In February 2026, a university student accused of publishing false information in an AI-generated image showing a funeral procession in alleged reference to President William Ruto was acquitted. The Milimani Law Courts held that prosecutors had failed to prove the student was responsible for publishing false information about President Ruto.

Constitutional challenge

In February 2026, activists filed a petition in the High Court at Kerugoya under Articles 22, 23 and 159 of the Kenyan Constitution, requesting that the Court immediately intervene to prevent the rollout of AI technologies they described as high-risk, citing potential violations of constitutional rights. The petitioners sought conservatory orders restraining the ICT Cabinet Secretary and Principal Secretary in the State Department for ICT from deploying, authorising, or operationalising AI systems pending the hearing and determination of the petition. They argued that AI systems are being rolled out without adequate legal, regulatory, or institutional safeguards.

In April 2026, the Court issued a ruling in Wangai & 2 others v Cabinet Secretary, Ministry of Information, Communications and the Digital Economy & 3 others [2026] KEHC 5690 (KLR), on the conservatory-orders motion. The Court declined the conservatory order, but on evidential rather than substantive grounds (para. 50). It accepted that the matter was urgent given the pace of AI deployment in employment, education, health and financial services and the approaching 2027 elections, and accepted that the rights violations alleged were likely dangers ‘if evidence of such deployment was adduced’ (paras. 35-36). However, the petition was missing particularity, as it failed to identify any specific AI system, place of deployment, and causal link between any identified system and the rights said to be threatened (para. 37). It ordered the petitioners to file an amended petition by 15 May 2026, set a hearing for 16-18 June 2026, and reserved judgment for 29 June 2026 (para. 50). No judgment had been reported as at September 2026.