Meta announced a content agreement with Newsmax to use its reporting and archives to train its AI models, raising concerns about misinformation.
Meta has struck a content‑licensing deal with Newsmax, the far‑right news outlet known for promoting conspiracy‑laden stories, to feed its reporting archives into the training data for the company’s next generation of artificial‑intelligence models.
What the agreement entails
Under the partnership, Meta will gain access to Newsmax’s articles, videos and historical archives, allowing its AI systems to learn from the outlet’s language patterns, topics and style. The move is part of Meta’s broader push to expand the data pool for large‑language models that power features like generative chatbots and content‑summarization tools.
Why Newsmax raises red flags
Newsmax has repeatedly published pieces that echo unfounded claims about election fraud, COVID‑19 misinformation and other conspiracy theories. Critics argue that incorporating such material could teach AI models to reproduce or amplify false narratives, undermining efforts to curb misinformation online.
- Newsmax’s coverage of the 2020 election included unverified allegations of voter fraud.
- The outlet has promoted debunked theories about COVID‑19 vaccines.
- Its editorial stance aligns closely with far‑right political figures.
Meta’s response and safeguards
Meta says the agreement includes strict content‑filtering protocols and that the AI will be trained to recognize and flag unreliable information. The company also claims it will use the data to improve factual accuracy and contextual understanding across its platforms.
We are committed to ensuring our AI systems are trained on high‑quality, diverse sources while implementing robust safety measures.
The announcement has sparked a debate among policy experts, who warn that even with filters, the sheer volume of biased content could seep into model outputs, making it harder to distinguish fact from fiction.
Yahoo coverage of Meta’s AI training partnership with Newsmax
Comments
No comments yet.