Experts Can't Agree How to Stop AI Lies, but Say Health Needs Different Fix

AI can now create fake videos, fake articles, and fake audio that look real. A new study asked 54 international experts to rank these threats in two areas: politics and health. The results show that the danger changes depending on the topic.

In politics, video deepfakes scored the highest threat level — almost 6.3 out of 7. A fake video of a politician saying something outrageous could spread in minutes. But in health, experts said AI-generated text was the bigger worry, scoring 5.8 out of 7. That means fake health advice, fake drug warnings, or fake news about a disease could be more harmful than doctored videos.

The experts also disagreed on how to fight back. Government regulation was named both the most effective solution (30% of votes) and the least effective (15%). That sounds contradictory, but the researchers say it's not a split between believers and skeptics — it's a disagreement over priorities. Some think rules are essential; others think they'd be too slow or hard to enforce.

The takeaway is simple: there's no single fix for AI disinformation. What works for political fakes may not work for health fakes. The study is an early map of a problem that's still growing, and it suggests we need a mix of tools — tech, education, and maybe regulation — tailored to each area.