A New Wild Cat, AI Disinformation, and Stress-Testing What We Think We Know

Scientists in Bolivia documented the first new wild cat species identified in approximately a century. Described in a peer-reviewed publication, the animal represents a distinct taxonomic unit from any previously known small cat lineage in South America. A century between confirmed new wild cat species discoveries reflects both how thoroughly the world's megafauna has been mapped and how exceptional genuine novelty has become. The conditions that allowed a distinct cat lineage to persist undetected — remote Andean and Amazonian habitat with limited human presence — are, researchers note, precisely the conditions most threatened by deforestation and climate change.

U.S. intelligence reportedly found that Russia is using AI deepfakes to target the 2026 midterms. The capability itself — AI-generated video and audio impersonating candidates, manufacturing false controversy, or suppressing turnout — is real and documented. Whether it will prove decisive is a harder question. Decades of political science research suggest voters are more resistant to persuasion through specific media exposures than propagandists typically assume, anchoring instead to party identity, economic conditions, and personal experience. The 2016 Russian influence campaign, which was real and extensively documented, is assessed in postmortems to have had modest measurable effects on vote totals.

For AI disinformation to have decisive electoral effects in November, it would need to reach genuinely persuadable voters — not already-committed partisans — at exactly the right moment, on concerns they were not already processing through other channels. Younger voters in particular have shown increasing sophistication about questioning the provenance of viral content, and AI-generated imagery has become easier to detect as media literacy around the artifacts of AI generation has improved faster than many anticipated.

A concrete signal to watch: if AI disinformation matters decisively, it should appear targeting narrow swing constituencies in races with margins under three points — not broad national audiences. If those close races track economic indicators and candidate approval as in every previous midterm cycle, that is evidence AI anxiety is outrunning AI impact. Watch also the ratio of voter-facing AI deepfakes flagged by platform moderation to engagement with that flagged content before removal. High flagging with low engagement suggests the immune system is working. High flagging with high engagement before takedown suggests otherwise.