Anthropic exposes AI operation that DRC journalists never knew existed
The Atlas Times automated desk · Published
Updated
Original reporting by Laura · Global Voices
Adapted source excerpt: CC BY 3.0; adapted summary Terms and license
Source published
70 AI-powered fake news sites have generated 318 articles on the eastern DRC conflict, relations with Rwanda, and contested mineral resources.
AI-assisted analysis. Published after automated checks; no human review.
Summary
In September 2026, AI development company Anthropic revealed a sophisticated disinformation network targeting the Democratic Republic of the Congo (DRC). The operation involved approximately 70 AI-powered websites that produced at least 318 articles focused on the conflict in eastern DRC, relations with Rwanda, and contested mineral resources. The campaign was orchestrated by the LKM Company, a French advertising agency, and amplified through coordinated accounts on X that mimicked legitimate media outlets. While no direct government involvement was identified, the scale of the operation alarmed African media watchdogs, particularly because it emerged during a period of low public trust in news and limited local resources. Local experts, including Hobab Watukalusu from Soma Média Lab in Kinshasa, noted that the speed and volume of AI-generated content outpaced traditional fact-checking methods. Journalists in the DRC face significant structural challenges, such as unreliable electricity, costly internet access, and high living expenses, which make it difficult to monitor industrial-scale disinformation while covering an active military conflict. Furthermore, the cost of analytical tools often limits the depth of verification possible in strained diplomatic environments. The impact of this disinformation extends beyond public opinion to tangible diplomatic consequences, including weakened DRC-Rwanda relations, restrictions on universities in M23-controlled areas, and entry limits announced by Burundi. In Rwanda, pro-DRC narratives portraying the country as a military occupier have deepened local divisions. However, the rise of AI disinformation also poses a risk to press freedom; the case of journalist Stanis Bujakera, who was arrested and sentenced for spreading false information between 2023 and 2024, shows how accusations of disinformation can be used to silence critics. To combat these trends, local initiatives like Kinshasa News Lab and partnerships with the International Fact-Checking Network (IFCN) and Meta are being leveraged, alongside free tools like InVID and Zero GPT. Despite these efforts, experts warn that training must evolve to help journalists understand the mechanics of generative AI and coordinated campaigns to maintain the credibility of fact-checking.
Key points
- Anthropic identified a network of approximately 70 AI-powered fake news sites that generated at least 318 articles concerning the eastern DRC conflict, Rwanda relations, and mineral resources in September 2026.
- The disinformation operation was attributed to the LKM Company, a French advertising agency, with coordinated amplification on X mimicking legitimate media, though no direct government involvement was found.
- Local fact-checkers in Kinshasa, such as Hobab Watukalusu, noted that the scale and speed of AI-generated content exceeded their previous detection capabilities, creating a significant resource gap.
- Journalists in the DRC face structural obstacles including unstable electricity, expensive internet, and high living costs, which limit their ability to analyze industrial-scale AI content during active military conflicts.
- The Stanis Bujakera case (2023-2024) illustrates how accusations of disinformation can be weaponized in legal proceedings, potentially stifling press freedom and complicating fact-checkers' work.
- Disinformation has tangible diplomatic impacts, including weakened DRC-Rwanda relations, specific restrictions on universities in M23-controlled areas, and entry restrictions announced by Burundi.
- Rwandan journalists report that pro-DRC disinformation portrays Rwanda as a military occupier, making it politically charged to distinguish between malicious narratives and legitimate criticism.
- Initiatives like Kinshasa News Lab and partnerships with the IFCN and Meta are helping combat false narratives, while free tools like InVID and Zero GPT are being utilized to detect AI content.
Context and interpretation
The emergence of industrial-scale AI disinformation in the DRC highlights a critical asymmetry in the information ecosystem: while AI can generate hundreds of targeted articles in days, traditional newsrooms require weeks for verification. This gap, combined with infrastructure deficits and the potential for disinformation to be weaponized against journalists, threatens to erode public trust and diplomatic stability in the region, making independent regulatory mechanisms essential to prevent anti-disinformation policies from becoming tools of repression.
Source: Global Voices