If I sell my snake oil as "fake snake oil", does that make it less deceptive? Or to a lesser degree, if my snake oil is so obviously fake that no normal person would misinterpret it as having any value whatsoever, does that make it less deceptive?
Watching the detectors: Researchers probe efficacy — and danger — of AI detection tools
May 2026, phys.org
- I wonder if ai wrote that title (not one but two em dashes)
- This is pretty damning on a whole industry that doesn't really work yet but is being forced into the market regardless.
- The writeup doesn't provide a link to the original article, and I suspect we're all being misled, and I don't mean by this article, I mean by the whole business.
Image credit: AI Art - Snake Oil - 2025
It only takes one fake web page to fool AI shopping bots, study finds
Jun 2026, phys.org
The researchers built a simulation tool called FORGE (Fake Online Recommendations in Generative Environments) to test 12 to evaluate web content pollution in AI models. They identified the main brand being discussed on selected pages and swapped it for a fake one. They did this for 225 products spanning 15 categories. After rewriting these pages, they tested whether LLMs would fall for the deception and include a fake brand in their recommendations.
"Across 12 commercial and open-weight LLMs, all models are vulnerable: a single polluted page yields fooled rates of up to 27%, while the full top-3 replacement raises this to 73.8%"
via UC Santa Cruz: Minghao Luo et al, One Polluted Page Is Enough: Evaluating Web Content Pollution in Generative Recommenders, arXiv (2026). DOI: 10.48550/arxiv.2606.13610
The consequences of relying on AI for accurate news
Jun 2026, phys.org
Participants who relied on AI systems to verify facts actually got worse at detecting misinformation on their own when their chatbots were taken away.
The authors also point out that the original human-created news content used to train the AI models is increasingly unreliable and/or biased, further exacerbating the problem.
Jesus we are f*cked
via MIT Media Lab: Anku Rani et al, Dialogues with AI Reduce Beliefs in Misinformation but Build No Lasting Discernment Skills, Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (2026). DOI: 10.1145/3772318.3790656
Polymarket's viral videos showed people winning big, but the bets were fake
Jun 2026, Ars Technica
“In its push to draw users to its unregulated platform, Polymarket has flooded social media with videos like Makihara’s, which appear genuine at first glance. In reality, Polymarket built near-perfect copies of its website, then instructed creators to make simulated trades on those dummy sites and hide that they were being paid by Polymarket.”
Makihara, a college student, posted a video in January “that showed him winning $100,000 on a wager that President Trump would publicly say the word ‘McDonald’s’ that month.” But trade data showed that no one on Polymarket won such a bet in January, according to the Journal. This was one of 145 bets that Makihara appeared to place on Polymarket between January and May, but all of those bets were fake, the article said.
“Many of the videos share a template: The creators open Polymarket, place a bet, and frequently refer to their winnings as ‘free money.’ Dozens of social-media creators have posted videos with almost identical formats. Polymarket sends creators bullet-point guidance on what to say, according to creators who have worked with the company and a recruiting website.”
via Wall Street Journal investigaiton (of all places)
No comments:
Post a Comment