In the heart of Kenya's vibrant media landscape, a quiet revolution is underway as Artificial Intelligence (AI) technologies progressively transform the traditional newsroom.
From content creation and distribution to audience engagement, AI-powered tools are reshaping how journalists operate, bringing with it unprecedented efficiency and innovation.
However, there is a pressing concern - the risk of media capture as algorithms and automation become integral to news production. Experts fear this technological revolution could be a double-edged sword of innovation and a new toolkit for media censorship.
Press freedom and standards of journalism are under heavy blows around the world - from media capture, where a deliberate attempt is made by powerful actors, including governments, to co-opt or control the media, silence journalists, and even shut down media outlets.
This was tragically illustrated in April 2015 when the Weekly Mirror's editor and publisher, John Kituyi, suffered a fatal blow to his head in Eldoret town, leading to permanent closure of the publication. The attack was believed to have been in response to the newspaper's critical reporting.
Media capture also occurs through government regulations, gag orders, civil suits with exorbitant damages, and denial of advertising which targets the sustainability of independent media.
Insidious challenge
However, AI-powered tools present a new and more insidious challenge. These have the potential to turn media organisations into unwitting accomplices in their own capture.
"Soft" media control is not new in Kenya. However, its intensity and visibility increased significantly from 2013 when the Jubilee government assumed power. Soft censorship through advertising and digital tools continues to be the biggest threat to media freedom, as it weakens both the financial stability and credibility of media organisations.
A 2021 report by ARTICLE 19 Eastern Africa found that both the State and corporate entities use financially induced self-censorship as a tactic, where they offer lucrative advertising and threaten to withdraw it if the media reports critically about them. These practices damage the foundations of the media industry and hinders the sustainability of independent outlets while discouraging collaboration.
Henry Maina, a media expert and former media Complaints Commissioner, says the risk of losing advertising revenue can influence editorial judgement, leading to decisions driven by factors other than news value. He contends that this not only compromises professionalism but also creates an environment where both public bodies, such as the Government Advertising Agency (GAA), and private marketing services entities, hold immense sway.
"When advertising departments become far bigger in decision-making, while editors take responsibility, this dynamic effectively turns media organisations into mere conduits for partisan reporting," Maina says.
Payment functions
The establishment of GAA in 2015 and the launch of MyGov, a government-owned publication, as a weekly pull-out in the four main newspapers in Kenya, increased the threat to the media's survival. MyGov consolidated procurement and payment functions of all advertisements. While these actions were framed as cost-cutting measures, promoters of press freedom saw them as deliberate attempts to either silence media critics or influence reporting. This move, combined with the GAA's influence, strengthened the capacity for soft media control and journalists' self-censorship. Government spending control is significant, as an estimated 30 per cent of newspaper revenues come from government advertising.
The Kenya Kwanza government has made the situation even more difficult, as leaders have made no secret of their contempt for the media, often accusing it of being "biased" or part of powerful "cartels" that need to be "crushed". One of its first moves after assuming office in September 2022 was to slash 75 per cent of the annual media advertising budget. All these add a new layer to media control, where risks extend beyond overt external pressures to subtle, less discernible influences of AI-driven tools such as algorithms and automation.
The use of algorithms to distribute online news content has raised concerns about transparency, objectivity and fairness in newsrooms. Algorithms generate news based on predicting individual preferences or based on an individual's past consumed information, which deviate from the journalistic principle of prioritising the public interest. To increase web traffic and attract big advertisers, some media outlets with an online presence are adopting clickbait-style language in their online content. Experts say clickbait may hurt online news coverage and circulation, leading readers or viewers to fall into 'filter bubbles' that reinforce their preconceived notions. They argue that not only does this raise questions about the impact on journalistic standards, but it also threatens public interest journalism.
Demas Kiprono, the Deputy Executive Director of the Kenya chapter of the International Commission of Jurists (ICJ-Kenya), says AI-powered tools can help journalists analyse large amounts of data and uncover hidden insights for in-depth investigative reporting. However, certain journalistic standards must be adhered to when using AI-driven tools to generate news articles and reporting. He points out the need to address issues such as algorithms with built-in biases or inaccuracies, deep fakes, and sensationalism influencing news curation rather than fairness and accuracy, which are cardinal ethical standards in journalism.
"There are concerns over automated creation of news articles that may need to meet specific standards, such as objectivity and authenticity. Media houses need to understand AI tools, their capabilities, limitations, and risks," says Kiprono
A study published in May 2023 by International Media Support titled 'AI, Journalism, and Public Interest Media in Africa', revealed the widespread use of AI in newsrooms across Africa. The study found that AI is being adopted in African newsrooms, with Machine Learning (ML) and Natural Language Processing (NLP) being the most common types. Newsrooms are using ML for fact-checking, verification, transcription, translation, data visualisation, sentiment analysis, and opinion mining.
Across newsrooms
The survey, which involved eight Kenyan media organisations, highlighted how differently AI tools are being adopted across newsrooms. Kenya and South Africa lead in AI use, particularly in well-resourced media organisations that invest in premium systems and custom-built tools. Smaller media organisations with limited resources have either not adopted AI or primarily use open-source tools.
From interviews with some editors, it was revealed that other AI tools being used in Kenyan newsrooms include Grammarly and Quillbot to detect errors and simplify or rephrase sentences. AI-powered Chatbots are being used to interact with audiences and gather feedback. Bing is used for keyword generation and Search Engine Optimisation (SEO) while ChatGPT is being used to analyse large amounts of text data, identify and verify key facts and figures, as well as help research and tease out news angles or headlines.
However, some editors are reluctant to use these tools directly, while others are uncomfortable with AI-driven tools in newsrooms. The use of AI technologies in journalism raises concerns about an increase in low-quality and polarising content, which could further erode public trust in journalism.
Michael Ollinga, Current Affairs and Special Reports Editor at Tuko.com, emphasises the need for caution when directly publishing AI-generated content. He explains that AI tools can act as catalysts, stimulating creative thinking and offering different approaches or angles to a story. His newsroom has monitoring tools that flag any fully AI-generated content against established codes. They monitor and counter content manipulation on social media platforms and search engines. They also regularly adjust their content strategies and encourage innovative content creation to overcome the limitations of AI tools such as hallucinations.
Ollinga says transparency is very important and they constantly remind journalists to interview experts from various fields to provide real and fresh perspectives for their stories. He says this is one of the steps taken to protect media freedom and journalistic standards from AI tools.
"To maintain editorial standards, we use monitoring tools that efficiently flag fully AI-generated content that may violate our established guidelines and insist on our journalists to interview experts in that field of their story topic," adds Ollinga.
Academic director
Prof George Ogola, a former lecturer and academic director of the MBA programme at Strathmore University Business School, believes AI tools will continue to bring about major benefits despite human anxieties of losing control to automation. The benefits include increasing people's span of control, reducing human error, and providing unprecedented efficiency.
"The technology still faces challenges arising from human self-interest, weaknesses, incapacity, and fear of losing control. These challenges do not serve the public interest. There is a need for collaboration between digital platforms and news media to improve technology and strong self-regulation to ensure unbiased public service," says Prof Ogola.
Press freedom ensures journalists can report independently without fear of interference from any entity, including government institutions, powerful individuals, or tech companies that power AI tools. Media experts express concerns regarding these technological businesses' profit-driven business model, concentrated power in only a few hands, lack of transparency, and potential control by a repressive state.
Prof George Nyabuga, of the Aga Khan University's Graduate School of Media and Communications, notes that most media consumers find it difficult to distinguish between AI-generated news articles and those written by humans.
"This digital media and information illiteracy means people can easily be deceived by increased misinformation and disinformation. This will ultimately affect people's trust in media and journalism, and therefore press freedom," he observes.
Privacy advocates have raised concerns about the risk of media control facilitated by advanced surveillance technologies. The government's covert online monitoring, often justified as combating disinformation, misinformation, hate speech, and national security-related issues under the Computer Misuse and Cybercrimes Act (CMCA) adopted in May 2018, has been highlighted as a potential threat to press freedom. This legislation includes provisions related to the publishing of "false information."
Press freedom advocates have warned about the potential for media control through these technologies, stemming from the government's covert online monitoring, often justified as combating disinformation, misinformation, hate speech, and other national security-related concerns under the Computer Misuse and Cybercrimes Act (CMCA) adopted in May 2018 to deal with "publishing of false information."
During the August 2022 elections, the Media Council of Kenya reported a surge in misinformation and disinformation campaigns across digital platforms, from media outlets to influential social media users. Foreign entities were also mentioned for using AI tools to spread online misinformation and disinformation in Kenya. Anonymous bloggers, allegedly associated with an Israeli 'Team Jorge' hacking group, allegedly shared fake documents online to discredit the credibility of the presidential election results.
The Guardian newspaper revealed that Team Jorge had previously been linked to disinformation efforts in the 2019 Senegal and the 2015 Nigerian presidential elections by accessing and sharing content from targeted social media accounts without users' knowledge or consent.
In 2017, Privacy International (PI) report disclosed that Kenya's national security agencies, particularly the National Intelligence Service (NIS), had direct access to communication systems, allowing interception of both traffic data and content. The report highlighted the presence of law enforcement and security agents in telecommunications facilities, facilitating unrestricted sharing of intercepted information among government agencies. Additionally, two NIS cybersecurity initiatives were scrutinized, raising concerns about their capability to monitor both content and traffic.
Governments world over are increasingly utilizing advanced AI-driven surveillance technologies to control social media and surveil dissenting voices, including the media. These technologies enable real-time monitoring of social media to identify any indications of opposition or negative coverage toward the government. Techniques involve the use of extensive datasets, keywords, and facial recognition to surveil government critics, including websites, bloggers, journalists, and content critical of leadership and governance styles. Repressive governments employ AI systems to suppress opposing political views, discredit critical online reporting, impose restrictions on major websites and social media, and manipulate or counter online public criticism with favourable narratives.
In August 2020, a website that questioned former President Uhuru Kenyatta's administration on the amount of money lost to corruption was taken down after the platform's owner, Charles Gichuki, was arrested and later released without charge. Authorities have also forced regular online users and journalists to delete content from their social media accounts and websites. On May 27, 2022, X (formerly Twitter) suspended 22 accounts, including those of the #NjaaRevolution campaign, protesting rising prices of foodstuff and cost of living in Kenya. The vague reasons, "suspicious behaviour" and "breaking X rules," raised concerns about potential state interference and the risks associated with AI tools in surveillance and opinion manipulation.
News consumers have also expressed concerns about the potential inherent bias in algorithms that could be exploited for political or commercial interests. Bias in algorithms can arise from various sources, including the data used to train them, the design of the algorithms, or the goals of the individuals or organisations developing them. This, they say, creates an environment where information is controlled and manipulated for specific agendas, leading to the suppression of diverse voices.
Others argue that media outlets are facing a dual challenge: the risk of being replaced by AI and their own political biases. Prof Bitange Ndemo, Kenya's ambassador to the Kingdom of Belgium, emphasises the need for capacity building in the development of algorithms. In 2018, Ndemo led an AI and Blockchain taskforce that recommended using these technologies to combat corruption and enhance transparency in Kenya. The report highlighted the dependability of their 'immutable' records. He states that there are efforts to create legislation and regulatory frameworks that specifically address the challenge of governing AI.
"Media itself is under threat of extinction from AI and is more biased now than ever before. We need capacity building in the development of algorithms. That way some of them can be discredited," says Prof Ndemo.
