August 1, 2026
Snapchat announced this week that its recommendation system will no longer reward fully AI-generated videos on Spotlight, TechCrunch reported. Only content made by real people will be eligible for the platform's discovery push going forward. This is a meaningful reversal for a company that, like most of its peers, spent the last two years courting AI-generated content to fill feeds cheaply. Now Snapchat is betting that authenticity, not volume, is what keeps users coming back. I think this is the right call, and I expect more platforms to follow. The economics of AI slop were always going to collapse once users started noticing the sameness -- the uncanny faces, the recycled templates, the content that technically fills a feed but says nothing. Snapchat's own Spotlight became a testing ground for that fatigue, and pulling the algorithmic reward lever is the fastest way to fix it without banning AI tools outright. For any business using AI to generate marketing content, the lesson is blunt: platforms are starting to penalize volume-for-volume's-sake, and a marketing strategy built purely on AI output at scale is now a liability, not an edge.
One day after launching, Google killed its Earth AI feature -- a tool that let anyone generate fake AI imagery and overlay it directly onto real Google Earth maps -- after backlash over how easily it could spread misinformation, according to TechCrunch. This wasn't a slow-burn controversy; it was a same-week reversal, which tells you Google either didn't stress-test the misuse case or badly underestimated it. What's striking is the contrast with Snapchat's move. Snapchat spent effort building guardrails around content it doesn't fully control. Google shipped a feature that made misinformation trivially easy to produce on its own flagship mapping product, and only pulled it after public pressure. That's not a technology problem -- it's a judgment problem, and it's the second time in recent memory that a major lab has had to walk back a launch because the obvious abuse case wasn't caught before ship. Businesses evaluating AI vendors should treat launch-day recalls like this as a signal worth watching: a company's internal review process for misuse is at least as important as its model quality. It's also a reminder that AI-generated content credibility is becoming a real differentiator, something we've touched on in The Web Is Choking on Its Own AI Output.
Apple CEO Tim Cook floated the idea that heavier Siri AI users could eventually pay more, likely bundled into iCloud+ subscriptions, per TechCrunch. Meanwhile, TechCrunch also reported that India's app market just posted a record $345 million quarter, with users increasingly willing to pay rather than just download for free. Put those two together and you get a clear direction: the free-AI-feature era is ending, and usage-based or tiered pricing is becoming the default, even for assistants baked into your phone. For businesses, this is worth watching closely if your own product roadmap assumes AI features are a free add-on that drives retention. Apple testing a paywall on its most consumer-facing AI product suggests even the biggest platforms don't see infinite-compute-for-free as sustainable. If you're building or buying AI-powered tools for your own operations -- whether that's a helpdesk with AI triage or a CRM with AI-assisted outreach -- expect vendors to start segmenting pricing around AI usage rather than seat count. That changes how you evaluate total cost of ownership, and it's part of why we've argued that understanding true costs before committing matters more than ever -- see our take in Buy vs. Build vs. ViibeStack.
Which of these worries you more as a business buyer: platforms quietly gating AI features behind new paywalls, or platforms shipping AI tools without adequately testing how they'll be misused?
Sources