Digg Reboot Fails Again

Digg tries again, this time as an AI news aggregator: In an email to beta testers, the company said the site's goal is to "track the most influential voices in a space" and to surface the news that's actually worth "paying attention to." https://lnkd.in/d5CZHcj8

Digg is back. For those who weren't online in 2008, Digg was one of the most visited sites on the internet — a social news aggregator where users upvoted stories to the front page. It got killed by a disastrous redesign in 2010, Reddit ate its audience, and it has been running a skeleton operation ever since. The 2026 version: an AI aggregator that monitors X for signals about which news is getting traction, surfaces the stories, and aggregates them. Human upvotes replaced by algorithm-detected engagement. The front page is curated by a model now. This is a reasonable product idea. It is also exactly the kind of thing that happens when a brand still has recognition but lost the core mechanic that made it work. The original value proposition was the crowd. The new version is a machine watching a different crowd and summarizing what it finds. Whether "Digg but AI" is a better news experience than "Twitter but worse" is what nobody knows yet. The name still works, though. #AI #Media

DuckDuckGo installations jumped in the last couple of weeks (some say 30% others 60%), and the business reports the highest ever search traffic. 30% from DuckDuckGo is not the same as 30% from Google, but is this a sign of a significant trend against "force-fed AI", or an anecdote?

Bots aren't traffic. They're distribution. 7 days of crawler activity on entia.systems: Googlebot: 1.62 GB GPTBot (OpenAI): 949 MB Meta-ExternalAgent: 498 MB ClaudeBot (Anthropic): 431 MB BingBot: 332 MB OAI-SearchBot, ChatGPT-User, PerplexityBot, Applebot, Amazonbot, Google-CloudVertexBot, ByteSpider, Manus… Every major LLM is reading the same verified entity registry. Most companies treat this traffic as cost. We treat it as the distribution channel that replaces SEO. The thesis behind ENTIA is simple: in an LLM-mediated economy, the entities that win aren't the loudest — they're the most legible. JSON-LD, government-anchored identifiers, eIDAS-signed provenance, zero cognitive resistance for the model. The action column on every row says Allow. That's not a bug. That's the product. (Verified business identity infrastructure for the AI era. 5.2M+ entities across 34 countries.)

I missed this last week somehow. An attacker used Meta's AI support for account takeover. While there are a few steps involved, the key was the AI's ability to add a new email address to an account without proper authorization. We don't know the details of the behind-the-scenes AI interaction, but there's a good bet there's an MCP server or other API allowing the agent broad access to edit accounts. https://lnkd.in/gzVJT9EH

Your engagement is fake, your followers aren’t real, the comments you read and argue with are auto-generated. And that was before we gave AI the ability to make profiles, images, posts and content on its own. It is called “dead-internet-theory”, and it is forcing us to change how we use the internet. Or we will be a victim of our own tools.

The Internet Has Crossed a Line For most of the history of the internet, there was an underlying assumption: there were people on the other side. People reading, searching, clicking, arguing, and sharing. Not anymore. New reporting indicates that automated bots and AI agents now generate more web traffic than human beings. Depending on the measurement, bots account for between 51% and 57% of all internet activity. For the first time, humans are no longer the majority users of the web. Ironically, we need AI to ensure our voice is heard.

Automated bot traffic has officially surpassed human internet activity. 57.5% of all web requests are now bots, according to Cloudflare. This surge is being driven by agentic AI - bots from OpenAI, Anthropic and Perplexity scanning the web constantly when responding to search prompts. If your brand's website and content isn't readable by this new wave of AI bots, you're unlikely to show up as a recommended option in ChatGPT, Claude, Perplexity or Google AI Overviews. The brands getting this right now will have a significant head start on those who don't.

Google last week unveiled its “biggest upgrade” to its search tool since the company first hit the web. It’s needed. So let’s hope it fixes Google’s flaws rather than worsens them. Google doesn’t want us to all start getting our digital info from ChatGPT, DuckDuckGo, Ecosia or Startpage. But, despite having access to some of the best minds in AI, Google’s first round of search results using AI left plenty to be desired. Story here: https://lnkd.in/gV2MXMjk

This is a reminder to not blindly trust whatever your favorite AI tool outputs. If you use LLMs on a daily basis or are actively working with agents or building agentic products - this is worth paying attention to What is this? - A Semrush study of ~150,000 LLM citations (June 2025) found that Reddit was cited 40.1% of time, Wikipedia 26.3% and YouTube 23.5%. This was across ChatGPT, Perplexity, Gemini, and Google AI Overviews. Together those three account for more cited weight than the next twenty domains combined. - The dominance is structural. Google licenses Reddit data for ~$60M/year, OpenAI signed its own deal in May 2024, and Perplexity manually boosts Reddit as a trusted domain alongside GitHub, Amazon, and LinkedIn. Why does it matter? - If you rely blindly on LLM output without cross checking or verifying or at least reading it yourself - you are subjecting yourself to biases of the users generating content on Reddit, Wikipedia and YouTube. - If you are using Agents - Observability, Evaluations and Governance are very important to have in place since you don't get to see each and every piece of output. There's a lot more to this story and what you can do to make sure you're using LLMs and AI Agents wisely. Check the link in the comment for a detailed rundown.