AGENCYwritten evidence (FON0017)

 

House of Lords Communications and Digital Select Committee inquiry:

The future of news: impartiality, trust and technology

 

 

  1. Introduction:

 

AGENCY is a multidisciplinary research consortium combining expertise in computer science (human-computer interaction, natural language processing, cybersecurity, and artificial intelligence), design, law, digital technology ethics, responsible innovation, social sciences and media studies. Members of AGENCY are researchers at the University of Birmingham, Durham University, Newcastle University, King's College London, Royal Holloway University of London, and the University of Surrey. A UK Research and Innovation (UKRI) grant supports our research as part of the Protecting Citizens Online programme. Grant title: AGENCY: Assuring Citizen Agency in a World with Complex Online Harms. Grant reference: EP/W032481/2.

 

This call for evidence into The Future of News: Impartiality, Trust and Technology coincides with the work, expertise, and concerns of the AGENCY project, which focuses on assuring citizen agency in a world with complex online harms. We refer to citizen agency as the ability of citizens (as users of digital technology) to exercise control over the use of (and access to) their data in the digital space. As such, we advocate for a citizen-centred approach to the design and implementation of digital technologies, recognising that regulation should premediate, mitigate, and respond to complex online harms by facilitating and supporting user agency and ensuring respect for fundamental principles such as freedom of expression. Given our expertise and commitment to these issues, we contribute to this discussion by hereby submitting our response to this call for evidence. Specifically, the following researchers contributed to the formulation of this response: Rebecca Owens, Viana Nijia Zhang, Dr. Shrikant Malviya, Dr Maksim Kalameyets, Prof. Abigail Durrant, Prof. Karen Elliott, Prof. Ben Farrand, Dr. Stamos Katsigiannis Dr. Cristina Neesham and Dr Lei Shi.

 

Our response incorporates insights from ongoing research, and we welcome any further inquiries.

 

  1. Executive Summary

 

This submission offers a nuanced socio-technical examination of the UK's evolving news landscape, analysing how online platforms can facilitate diverse voices while addressing challenges like mis- and disinformation. It advocates for innovative regulatory interventions, fostering transparency (particularly regarding algorithms), implementing due diligence, conducting risk assessments, and enhancing reporting protocols, all under Ofcom's guidance. In our view, such regulatory interventions should be supplemented by the adoption of a Corporate Digital Responsibility (CDR) framework to promote trust in technology platforms. Ultimately, our recommendations aim to resolve the news industry's complex and evolving challenges, ensuring a balanced, trustworthy information environment that protects freedom of speech.

 

  1. Trends over the next 12 months and 5 years

 

1. What impacts (positive and negative) do large technology platforms and online news aggregators have on the UK’s news environment, including media plurality? And how might this change?

 

Positive impacts:

 

  1. Increased accessibility: Technology platforms and news aggregators grant access to extensive news sources and viewpoints, broadening public access to information. This accessibility fosters a more informed citizenry, enriching public discourse and understanding.

 

  1. Citizen journalism and user-generated content: Social media platforms allow ordinary citizens to participate in news reporting and dissemination by sharing eyewitness accounts, photos, and videos of breaking news events. Platforms like Facebook and X (formerly Twitter) continue to play a crucial role in amplifying user-generated content during major news events such as natural disasters, protests, and elections, raising the visibility and influence of the lay (non-expert) reporter.

 

  1. Innovation in news delivery: Recent startups, like the World News Index and LettsNews, drive innovation in news delivery by offering features such as personalised news feeds and democratising access to the news by collating and curating global stories from multiple sources. The BBC News website has also implemented personalised news recommendations based on user’s browsing history and preferences to enhance their experience and engagement with news content.

 

  1. More support for independent journalism: Online platforms and community forums provide a platform for independent journalists and smaller news organisations to reach a wider audience without the need for significant financial resources. For example, platforms like Substack and Medium allow independent journalists to monetise their content directly through subscriptions, reducing their dependence on traditional media outlets.

 

  1. Increased media plurality: Platforms like Google News and Flipboard aggregate news from various sources, including mainstream media outlets, niche publications, and independent journalists, allowing users to discover diverse perspectives and news sources they may not have encountered otherwise, thereby contributing to a pluralistic and more diverse media landscape.

 

 

Negative impacts:

 

  1. Information overload, encounter, and avoidance: Large technology platforms leverage Artificial Intelligence (AI)-powered algorithms to expedite the creation and distribution of information. These algorithms assemble user data into undisclosed profiles, which are then utilised to subtly influence user decisions regarding products and services that third parties may benefit from. By analysing user behaviours, preferences, and interactions, these algorithms curate and present personalised, data-driven content and service offers. This mechanism of algorithmic curation, or 'feed', facilitates 'information encountering', allowing users to stumble upon information unexpectedly. This method diminishes the likelihood for individuals to validate the veracity and precision of information and seek out information actively.

 

In addition, this abundance of information may also adversely impact the public’s well-being. For instance, research around information overload from the Reuters Institute Digital News Report in 2019 showed that 28% of the respondents stated that they received too much news and felt worn out by it (source). In 2016, the Pew Research Center also found that 59% were worn out by the amount of news; in 2019, 66% made the same claim (source). Similarly, research has found that the use of online news and social media (source) (source) and specific news formats, for instance, push notifications (source), increased news overload for users. Conversely, the passive consumption of television news appears to have a lesser effect on users' perceptions of being overwhelmed by information. (source).

 

  1. Creation of echo chambers: The use of automated aggregation in combination with automated translation and potentially AI-generated content has also contributed to the creation of echo chambers, namely, contexts where one is exposed only to similar views and beliefs without any opportunity to consider alternatives (source). This is because the same news is disseminated in numerous news aggregators with minimal curation, leading to a false sense of importance and validity, which is further reinforced by search engines promoting such news due to the sheer amount of news aggregators that repeat them.

 

  1. Increased copyright infringement: Unauthorised use and dissemination of copyrighted materials could violate the intellectual property rights of the Original Content Creators (OCCs), impacting journalists both financially and reputationally. Furthermore, the increased use of ChatGPT and similar text-generation tools may reinforce the frequency of such infringements. Trends suggest that these tools are increasingly used to write news pieces, posing a greater danger of potential copyright violations.

 

  1. Misuse of AI technologies in content creation: The advancement of AI tools, such as Large Language Models (LLMs), has revolutionised content generation, introducing innovative methodologies that nurture an environment which promotes diverse creative expression. However, instances have emerged where news organisations have employed AI to produce articles that may be misleading or contain inaccuracies. For example, as of the 6th of February 2024, Newsguard has identified 687 AI-generated news and information sites operating with little to no human oversight (source). The presence of such sites displaying fake news could harm the integrity of information disseminated to the public and overall trust in the media.

 

  1. Disproportionate financial incentives: The rise of major technology platforms and online news aggregators has commodified news, leading to lower financial returns for journalists, writers, and publishers, and to disproportionately higher profits for search engines and social media platforms. Consequently, the creators of original news content may be inadequately compensated relative to the revenue generated by these technology platforms and may be ultimately displaced by technological innovation.

 

  1. Disinformation and misinformation: Tech platforms have been criticised for their role in spreading disinformation and fake news (source). The viral nature of social media can amplify false narratives and undermine trust in sources of information that are, in fact, reliable. Studies have shown that exposure to misinformation on social media can lead to the dissemination of inaccurate beliefs and can undermine public trust in institutions such as the NHS and the media (source) (source). Disinformation takes numerous forms, including using real journalist details to create fake accounts, creating fake journalists affiliated with diverse news services, or individual ‘citizen reporting’. Furthermore, dis-/mis-information can appear in the form of political advertising, which can be created by state or private actors and presented as ‘authentic’ news content or reference news content that may be (later) disproved.

 

  1. Risk of manipulation in aggregation algorithms. There is a potential risk associated with manipulating the news order within algorithmic aggregators, which undermines the importance of news and the probability of users' accessibility to genuine articles and items. Certain technologies and aggregation algorithms are vulnerable to cyber-attacks and social engineering tactics. Cyber-attacks could, for instance, reduce the ranking of specific news items, effectively hiding them from consumption by a wider audience.

 

Potential Changes:

 

        Regulatory intervention: The UK government has taken steps towards the introduction of social media regulations to address the spread of mis- and disinformation in the form of the proposed Online Harms White Paper/Bill. However, the journey of the Online Safety Act through parliament has resulted in the removal of obligations for platforms' consideration of disinformation in the context of broader service risk assessments. Implementing a legislative obligation on platforms to consider the risk of their services being used to spread disinformation, whether in the form of news content, individual social media posts or advertising can help address this concern.

 

Proposals to disclose the sources of political advertising, such as in the Digital Services Act and EU Proposal for a Regulation on Political Advertising, would be welcomed. The proposal imposes obligations of transparency, traceability and due diligence upon platforms and prohibits the use of micro-targeted advertisements that involve processing personal data, which can be used to tailor misleading political messages to individual users (source) (source).

 

In addition, increased transparency is necessary regarding the algorithms used to aggregate news. Algorithms shape users’ content based on arbitrary profiles created from prior searches, data scraping tools, etc. This may prioritise certain news items over others in ways that potentially expose users to misinformation or limit their access to diverse viewpoints. Society must understand how these algorithms function and address an array of concerns about fairness, biases in news ranking, and identification of any vulnerable components that external actors could exploit to manipulate users. In this context, corporate digital responsibility (CDR) frameworks can be applied as tools to assign important responsibilities to technology development and business decision-makers (e.g. who is responsible for transparency in algorithm design, negative impacts on users, etc.).

 

        Responsibility/revenue-sharing agreements: To mitigate disproportionate financial returns, news publishers and technology platforms could negotiate responsibility for safeguarding and revenue-sharing agreements to ensure journalists, writers, and publishers receive fair recompense for their content contributions. For instance, Google and Facebook could allocate a portion of their advertising revenue to support journalism through initiatives such as the Facebook Journalism Project. Similarly, a model adopted for music creators by Spotify could provide a potential solution to this problem, which can be replicated or adapted to similar online information-sharing contexts.

 

 

2. How is generative AI affecting news media business models, and how might this evolve?

 

In our view, we see it evolving in several key ways:

 

  1. Advanced news aggregators: As generative AI matures, multiple online Media platforms and news aggregators emerge (e.g., LettsNews, Briefly), which take high-quality journalism as input and convert the content feeds into either self-publishing (X, substack, Instagram, Facebook, WordPress, medium or others) or other media channel outlets (newspapers, e-papers, TV, radio and indi-channels like YouTube and others). This might decrease trust in the media as it may obscure the original sources of information, making it difficult for readers to assess the credibility of the content.

 

  1. Personalised content delivery: Generative AI algorithms can analyse user preferences and behaviour to deliver personalised news content, thus enhancing user engagement and satisfaction. While personalised content delivery offers many benefits, potential drawbacks include concerns over the use and re-use of personal data, algorithmically produced filter bubbles, a lack of data transparency in their functions, selling data to undisclosed third parties, a lack of data security, and the risk of promoting potentially harmful content.

 

  1. Chatbots for news delivery: Chatbots powered by generative AI provide users with a conversational and interactive way to consume news. Chatbots enable two-way interaction between the user and news provider to permit feedback, questions, opinions, and suggestions. Furthermore, they leverage artificial intelligence, natural language processing (NLP), and data analysis to create dynamic and intelligent content such as summaries, updates, alerts, and recommendations that add value and diversity to news offerings.

 

  1. Redefining journalistic roles: The advent of AI in automated news writing is reshaping journalistic roles, shifting journalists from routine content creation to analytical and investigative tasks. The collaborative workflow involves leveraging AI for data analysis, routine reporting, and content generation, freeing up journalists to focus on high-value tasks such as in-depth analysis, interviews, and investigative journalism.

 

  1. Pollution with rewritten news: Generative AI enables the rewriting of existing news in a manner that can evade detection as plagiarism. With the rise of generative AI, news companies require enhanced copyright and author rights protection to ensure journalists can be confident that their work will not be utilised without permission.

 

3. How are perceptions of due impartiality evolving, and what challenges do news organisations face around impartial reporting?

 

In our view, there are three key challenges:

 

  1. The ‘misreporting’ challenge: The impact of mis- and disinformation cannot be ignored, as it may influence public perceptions and undermine trust in content ‘experts’ and ‘professionals’. We must foster transparent reporting mechanisms and introduce supervision and review systems operated by both ‘gatekeepers’ such as journalists and pre-trained AI algorithms. One such approach that the government could adapt for the UK market is that of ForHumanity, which has designed an independent auditing procedure for AI Systems to safeguard users, promote trust and establish best practices to develop trustworthy algorithms (source).

 

  1. The information avoidance challenge: News organisations struggle to engage the younger generation (Gen Z onwards), who are more likely to source news information from informal social media sources (source). This cohort can intentionally or unintentionally hide from traditional news sources to avoid mood affection after consuming ‘overwhelming’ political or societal news content (source). As such, news organisations must adapt their strategies to engage younger audiences more effectively. One way this could be done is through presenting the news in communication styles inspired by those generated by and for young audiences on social media platforms.

 

  1. The ‘Deepfake’ challenge: Deepfakes require acute attention and effort to be promptly identified and removed from social media platforms to prevent the spread of mis- and disinformation and preserve the integrity of the online environment. One approach to tackling this problem may be the clear labelling of AI-generated material by adopting a mechanism like the European Union's forthcoming AI Act, which mandates that AI-manipulated content, such as deepfakes, carry a visible label disclosing the use of AI.

 

4. What factors affect trust in the news, and how might this evolve?

 

b) What impact do concerns around disinformation have on trust in the information environment? (To what extent does this differ between different sections of society?)

 

In answering this question, we would like to highlight the impact of disinformation on trust in recognised healthcare service providers.

 

 

  1. Evaluation

 

4. How well is regulatory oversight working? Are any changes needed, for example:

 

 

a) In the way Ofcom oversees due impartiality and the extent of its remit?

 

        Increase transparency of complaints procedure: To boost public trust in media oversight, Ofcom should increase transparency in its complaints procedure, being clearer about how it categorises and investigates public complaints concerning the relevant code (source). This lack of transparency risks eroding public confidence in the fairness and objectivity of media regulation.

 

b) In the way Ofcom oversees media plurality?

 

        More accountability for the lack of diversity within public service broadcasters (PSB): Ofcom has faced significant criticism for inadequately enforcing the BBC's charter commitments regarding workplace diversity (source). Further work is needed to ensure a more diverse workforce and to enable the creation of pluralistic content that accurately represents society. Similarly, the development of General Artificial Intelligence (GAI) predominantly involves white males from middle-class backgrounds, leading to potential implicit bias in these tools (source). Recognising that a lack of diversity may permeate different stages of the news generation cycle, the government should implement more holistic initiatives to promote diversity.

 

5. Are there any actions the Government should take to address concerns around due impartiality, trust, and the influence of technology platforms? The key to increasing impartiality and trust in technology platforms rests on increased commitments of all parties (news and platform providers, regulators, academics and policymakers) to transparency and the responsible development and use of data and digital technologies. To this end, we advocate that the government should:

 

        Establish a legal foundation for data accessibility: Article 40 of the European Union's Digital Services Act 2022 establishes a clear mechanism permitting vetted researchers to access data from exceptionally large platforms and search engines. This provision must be mirrored within UK legislation, thereby granting researchers access to crucial data through which the reliability and integrity of online platforms can be evaluated. Specifically, researchers should have the capability to extract current information from social media platforms through readily available Application Programming Interfaces (APIs). This would enable a more comprehensive understanding of the content that users are consuming, producing, and disseminating.

 

        Create an independent oversight body to audit and verify LLMs' training data and outputs: The training material of novel LLMs should be pluralistic, non-discriminatory and verifiable by robust and independent procedures. Safeguards ensure that the models produce accurate and unbiased information aligned with transparency and fairness principles advocated in the government’s “AI: A pro-regulation approach” white paper (source). Internationally, organisations such as ForHumanity (source) are already performing this role, and the government should learn from these organisations to develop best practices that align with the UK market and protect users.

 

        Codify the Digital Regulation Cooperation Forum’s (DRCF) mandate: Presently, membership of the DRCF is voluntary, which limits its effectiveness. Formalising DRCF’s mandate through statutory duties to collaborate, consult, and jointly regulate overlapping areas would serve to share resources and information and to make regulating technology platforms more efficient (source).

 

        Establish a legal right for users to know that they are interacting with chatbots: The forthcoming EU AI Act enshrines users’ right to know if they are interacting with chatbots, permitting critical assessment and avoidance of misinformation. This approach to facilitating and supporting user agency aligns with the government's call for increased transparency in AI applications, as outlined in its 'AI: A pro-regulation approach' white paper (source), and should, therefore, be implemented.

 

        Support the adoption of a Corporate Digital Responsibility (CDR) framework for technology platforms (source): CDR examines all of the issues raised in this document from the perspective of responsibility to avoid unintended consequences arising from the misuse of data and digital technologies, from conception and design phase through to distribution of content, product, or service. In short, it refers to “a set of practices and behaviours that help an organisation use data and digital technologies in ways that are perceived as socially, economically, and environmentally responsible” (source). This malleable tool provides any organisation or decision-maker with a frame of reference to understand risks in the digital age. Specifically, the protection and building of user trust sits at the heart of the CDR framework, warranting consideration and incorporation of responsibility principles at each stage of developing data-driven technologies. Based on the body of literature forming around CDR and practical global implementation (cf. ibid.), issues of trust are eminently socio-technical in that humans create the data upon which machine learning tools are trained and aggregated to AI (and now GAI). Trust, transparency, and maintenance of safeguarding users cannot emerge if accountability and responsibility are ill-defined and result in avoiding questions of legal culpability and/or liability when online harm occurs. The application of CDR frameworks is, therefore, a pivotal recommendation - as such tools should guide critical decision-making processes to cascade throughout the hierarchy of news and social media platforms. We argue that policing and compliance aspects of CDR, as per any white paper or Act, should include consideration of legal punishment for non-compliance by organisations or individuals. Legal means to enforce CDR principles, as opposed to voluntary adoption, would prevent CDR from becoming ‘washed’ - as has been the case, for instance, with falsely reporting targets in relation to meeting United Nations ESG directives.

a) Are changes needed to the Media Bill?[1]

 

        In our opinion, the government must introduce further measures to ensure that the public is at the centre of overseeing Public Service Broadcasters (PSBs): s5(b) of the Media Bill requires Ofcom to ensure “sufficient quantity of audiovisual content that reflects the lives and concerns of different communities and cultural interests and traditions within the United Kingdom.” Such requirements mandate active engagement by Ofcom with a broad spectrum of audiences to ensure that the content produced is representative and inclusive of the general public.

 

        Disclosure of the algorithms used for news aggregation: such regulations would provide transparency regarding how news items are selected, ranked, and displayed to users. This disclosure would promote accountability, enhance user trust, and mitigate potential risks of bias or manipulation.

 

b) Are changes needed to the way the Government addresses mis- and disinformation? From a regulatory point of view, the government should adopt a human-centred and stakeholder-centred approach to combating mis- and disinformation. In this context, we advocate for:

 

        Strengthened procedures to remove false information: The government should adopt a more holistic regulatory framework aimed at the meticulous removal of demonstrably false information. This comprehensive strategy should encompass the formulation of precise CDR guidelines, co-created with key stakeholders such as the media, technology platforms, and the general public, to identify, report, and remove false information in a timely manner.

 

        Improve cross-sectoral collaboration: The government should establish robust mechanisms for cooperation between platforms, regulatory bodies, and the wider academic community to support the development of best practices and the sharing of expertise and up-to-date information on risks posed by mis- and disinformation.

 

        Regulation to improve media literacy: Improving media literacy standards is key to providing users with the agency to navigate mis- and disinformation. Accordingly, it has been recognised by the government in the Media Bill and Online Safety Act. However, the measures delineated by the government are limited and should be supplemented by a continuous commitment to improving young people's media literacy through:

 

     Requiring schools and education establishments to have a media literacy policy: By mandating that schools and educational establishments have a clear media literacy policy, we can ensure that young people are equipped with the skills to identify mis- and disinformation and have the agency to engage with digital information safely and responsibly.

 

     Incorporating media literacy into the national curriculum: By incorporating media literacy into the national curriculum, the government can show a clear commitment to equipping young people with the critical thinking skills necessary to identify and/or bias and to evaluate the reliability of media sources. This approach would supplement the Essential Data Skills Framework (source) by equipping citizens with the skills to become digital citizens.

 

        Finally, to reiterate our key points, our recommendations are less about individual posts or behaviour and more about addressing the conditions in which ecosystems of distrust tend to be created, leading to generalised impacts of mis- and disinformation on wider society. In this context, traditional news and broadcast media can also be sources of mis- and disinformation. For this reason, a broader approach is required, to incorporate obligations of transparency (including algorithmic transparency), due diligence, risk assessment, and reporting, all overseen by Ofcom.

 

 

February 2024

11

 


[1]              CDR is a mechanism to underpin and deliver change.