This is an outcome report from the Trustworthy by Design Convening: AI, Content Authenticity, and the Future of Reliable Information. A gathering hosted by WITNESS and the Forum on Information and Democracy (FID) at the headquarters of the International Federation of Red Cross and Red Crescent Societies (IFRC) in Geneva on 8 July, 2026.
On the sidelines of the Global Dialogue on AI Governance, the World Summit on the Information Society (WSIS) and the AI for Good Summit, WITNESS and the Forum on Information and Democracy brought together a deliberately cross-sector group at IFRC headquarters in Geneva to explore a shared agenda for trustworthy information in the age of AI.
Around the table were people from government, humanitarian organisations, civil society, news media, standards bodies and technology companies, each working on the erosion of trust in digital content from a different angle, and rarely in the same room at the same time. The premise was simple: so much of the work on content authentication, AI transparency and information integrity is happening in parallel, with limited visibility across efforts, that there is real value in stepping back to map where it intersects, where it diverges, and where working together could amplify impact.
The opening session was anchored by its three host organisations. IFRC Secretary General, Jagan Chapagain opened by welcoming partners to the Federation’s house. In his remarks, Jagan made it clear that for the IFRC, questions of AI governance, information integrity and digital trust are deeply operational and central to the organisation’s humanitarian mission. WITNESS’ Associate Director, Mahsa Alimardani, gave the opening remarks and set the premise on which the day would rest: In her opening, she emphasised that there is no trustworthy AI without a trustworthy information environment.
“There is no trustworthy AI without a trustworthy information environment.”
Mahsa Alimardani WITNESS
WITNESS’ Bruna Martins dos Santos, moderated the first panel. During the session, Jacobo Castellanos, launched a new WITNESS report of which he was the lead author titled “C2PA Content Credentials and the Surveillance Risk: Adversarial Scenarios and Governance Gaps in the Content Provenance Ecosystem”. The second panel was moderated by FID Executive Director Camille Grenier, who also delivered the event’s closing remarks.
The convening was held under the Chatham House Rule. Accordingly, the reflections and key takeaways presented below capture the substance of the discussions without attributing specific comments to individuals participants beyond the host organisations.
Key takeaways
- The erosion of trust is a shared reality across every sector, not an open question.
It shows up differently for governments and their constituents, for humanitarian actors and the communities they serve, for newsrooms and their audiences, and for the civil-society organisations closest to those most at risk, but it is the same underlying condition. As Mahsa Alimardani put it, quoting the co-chairs of the UN’s first Independent International Scientific Panel on AI report, information integrity has become “the mother of all the battles”, and the window to act together is open now and may not stay open long. WITNESS sees both sides of the problem directly, from synthetic media targeting individuals to the liar’s dividend, where real footage of real atrocities is waved away as fake, a denial that has become a dominant threat in the conflicts WITNESS documents.
- In a crisis, information is a life-saving infrastructure.
Secretary General Jagan Chapagain drew on the IFRC’s World Disasters Report to show what is at stake, noting that false, misleading or manipulated information can obstruct humanitarian access, deepen fear, endanger volunteers and prevent people from receiving assistance. “In every crisis I have witnessed,” he said, “information is as essential as food, water and shelter.” The wider discussion reinforced that what makes this moment different is its scale, speed and systemic nature, with disinformation now operating as an industry, that could be better understood as a problem of freedom of reach than of freedom of speech.
- Provenance and authenticity infrastructure is necessary, but it carries a surveillance risk that has to be governed.
The centrepiece of the morning was the launch of a new WITNESS report, C2PA Content Credentials and the Surveillance Risk: Adversarial Scenarios and Governance Gaps in the Content Provenance Ecosystem. Jacobo Castellanos framed the opportunity as building infrastructure that does “two things at once”, helping to authenticate content so we can reclaim trust, and protecting the people who do that work while they do it. The report shows how the same features that make content credentials valuable, their metadata trails, signed certificates and identity assertions, can become tools for identity disclosure and expression control when deployed without safeguards, with journalists, human rights defenders and documentary filmmakers among the most exposed. Grounded in reality and reviewed by around thirty experts, it sets out seven adversarial scenarios and three pathways to harm (regulatory misuse, architectural capture, and implementation failures), and a roadmap of red lines and governance mechanisms to be built now rather than retrofitted after harm occurs. The standard Mahsa Alimardani asked the room to hold all day was blunt: “authenticity infrastructure cannot become surveillance infrastructure.” The report is available at wit.to/C2PA-privacy.
- Technical fixes alone are not enough; governance innovation is the lever we underuse.
As Jacobo Castellanos argued, of everything the report points to, governance innovation “may be the most impactful lever we have, and it is also the one that we spend the least time on”. The mechanisms this moment calls for do not fully exist yet, and building them will mean working across sectoral, disciplinary and institutional boundaries rather than assuming inherited tools will stretch to fit.
“Governance innovation may be the most impactful lever we have, and it is also the one we spend the least time on.”
Jacobo Castellanos, WITNESS
- The problem is one of power and economics, not only content.
Discussion returned repeatedly to the monetisation of disinformation, the concentration of market power, and the instruments that sit beyond any single AI law, from competition and antitrust to copyright and the fair remuneration of the information, culture and knowledge that feed generative AI. Alongside this ran a shared concern for the sustainability of public-interest journalism and the strain on the knowledge commons, as freely licensed content is scraped at scale without support for the infrastructure that maintains it.
Learnings and looking ahead
The clearest learning was one that came up throughout the sessions: technical solutions alone will not be enough to strengthen trust in the digital information ecosystem. This is a collective problem that needs collective action. The value of a convening like this lies precisely in creating the space, and assembling the range of people, to build governance that does not yet exist. Bruna Martins dos Santos pressed the group on exactly this point, asking who bears the cost when trust in information collapses, and urging participants to “draw a line through” the many parallel efforts rather than each “creating little wheels” in isolation.
Part of what made the convening necessary was a frank reading of the Global Dialogue on AI it sat alongside. WITNESS welcomed the prominence the Dialogue gave to information integrity, but was candid that naming harms is not the same as governing them: the Dialogue brought only a limited set of civil-society voices to the table, some major frontier developers stayed away, and Geneva was left without clear alignment between governments, companies and civil society, or a path forward grounded in rights. Closing that gap, between naming harms and governing them, and between consultation and genuine cross-sector collaboration, is precisely the work a convening like this one exists to advance.
Camille Grenier closed the convening by drawing the threads together and framing the day not as a conclusion but as a starting point, one built on the momentum and the fears that emerged from the Global Dialogue, and pointing FID and its partners towards the continuing, cross-sector work ahead.
“We will be reflecting on the spaces we have tried to support over the years, bridging AI, information and the future of media and journalism around the world.”
Camille Grenier, Forum on Information and Democracy
Several concrete threads are worth carrying forward: folding information integrity into competition and copyright remedies; sustaining public-interest journalism and the knowledge commons; and, above all, the governance innovation that content provenance urgently needs. Running through all of them is Jagan Chapagain’s request that the humanitarian perspective, the perspective of the communities most affected by harmful information, be “not only heard, but placed at the heart of discussions and decisions”.
Two invitations follow from the session:
The first is practical: to engage with the new WITNESS report and to explore how each sector can shape content provenance and the C2PA as these technologies mature.
The second is broader: to keep building together and across sectors, the mechanisms and the trust that no single organisation can create alone. This convening was a step towards acting more collectively, and towards the 2027 AI Impact Summit that will return to Geneva.
WITNESS and FID are grateful to the IFRC for hosting, and to everyone who joined the conversation. It is one we intend to keep going, together.