In a post on 9 October 2025, the eve of his 41st birthday, Telegram’s founder Pavel Durov wrote that “our generation is running out of time to save the free Internet built for us by our fathers”. The line has been repeated, shortened and reposted ever since, until “the open internet is dying” now turns up in everything from privacy campaigns to marketing newsletters.
The people saying it do not mean the same thing. At least three separate stories sit under the one phrase, each with its own evidence and its own people arguing either side. For a business with a website, the one most likely to matter first is the least political of the three, and you can already see it in your own analytics.
Who is saying it, and what they mean
The phrase is doing three jobs at once. Taking them one at a time is the only way to judge any of them.
1. Governments are tightening control
This is Durov’s meaning. His post listed “dystopian measures such as digital IDs (UK), online age checks (Australia), and mass scanning of private messages (EU)”. He also said Germany was persecuting anyone who criticised officials online, that “the UK is imprisoning thousands for their tweets”, and that France was criminally investigating tech leaders. “A dark, dystopian world is approaching fast,” he wrote. Durov has his own stake in the argument. On 28 August 2024 French judicial authorities placed him under formal investigation over offences committed on Telegram, and he has called charging a CEO with crimes committed by third parties on his platform “a misguided approach”.
Supporters of the new rules describe the same changes as overdue child protection. In the UK, pornography sites have had to use highly effective age checks under the Online Safety Act since 25 July 2025. On 28 August 2026 Ofcom said that in the Act’s first 18 months it had investigated more than 100 platforms and issued fines totalling more than £6m against 10 providers. It added that a further seven sites posing a significant risk of harm to children had blocked access to people in the UK after engagement with its supervision team. “Formal enforcement action is one tool we use to bring companies into compliance,” Ofcom said. In Australia, platforms have had to take reasonable steps since 10 December 2025 to stop under-16s creating or keeping social media accounts.
Whether this is protection or overreach is a political judgement, and reasonable people disagree. What is not in dispute is that the rules are now being enforced.
2. Big platforms and AI companies are closing it up
The second meaning has nothing to do with governments. It is the argument that a handful of companies now control how most people reach information, and that AI has made it worse. On a Tech Policy Press podcast titled “The Open Internet is Dead. What Comes Next?” in October 2025, the technologist Mallory Knodel put it bluntly: “There is no incentive to interoperate when it comes to AI, none.”
Google’s own statements point in two directions. In a filing in the US ad tech antitrust case on 5 September 2025, Google’s lawyers wrote that “today, the open web is already in rapid decline” and that the Justice Department’s proposal “would only accelerate that decline”. The filing was about open-web display advertising, the income publishers earn from ads on their own sites. In May 2025, Google’s Nick Fox told the AI Inside podcast that “from our point of view, the web is thriving”, adding that the death of the web “has been coming for 25 years. It’s not happening.”
3. AI answers and bots are replacing visitors
The third meaning is the most measurable. People are getting answers without visiting websites, and a growing share of what does visit is not human.
Pew Research tracked 900 US adults through 68,879 Google searches in March 2025. On visits where an AI summary appeared, people clicked a traditional search result 8% of the time, against 15% when there was none, and clicked a link inside the summary on just 1% of visits. Google told reporters the study’s methodology was flawed. SparkToro, using Similarweb clickstream panel data, found that 68% of US Google searches from January to April 2026 ended without a click, up from 60% in 2024.
Bots are the other half. Thales’s 2026 Bad Bot Report found automated traffic made up more than 53% of all web traffic in 2025, with bad bots alone at 40% of all traffic. Thales sells bot protection, and the figures cover the traffic it can detect. The companies in the middle are taking sides too. In July 2025 Cloudflare announced it would block AI crawlers by default unless they pay creators. On 1 July 2026 it said that from 15 September, every new domain joining Cloudflare would block AI training and agent crawlers by default on pages that show ads, while still allowing search crawlers.
The one already in your analytics
The first two stories are real, but most UK businesses will feel them indirectly, if at all. The third changes the numbers on your own website now, whatever you think of the politics. Three places to look.
Search clicks. Open Google Search Console and compare impressions with clicks over the last 12 months. If your pages are still being shown but fewer people click, search your main queries and see whether an AI summary now appears: Pew found clicks roughly halve when one does. A summary can explain a topic, but it cannot quote your price or take a booking, so those pages are less exposed than general explainers.
Who your visitors are. Google Analytics already excludes known bots automatically, and that cannot be switched off, so the bots that do get counted are the ones it does not recognise. Look at your form submissions too: a rise in junk enquiries is one sign. Tools such as Cloudflare Turnstile can stop form spam without asking customers to solve puzzles.
Whether AI crawlers can read your site. This is now a decision, and there is no single right answer. Letting AI crawlers in means your business can be described and cited when someone asks an AI assistant for a recommendation. Blocking them keeps your content out of AI training and answers. If your site joined Cloudflare on or after 15 September 2026 and shows ads, AI training and agent crawlers are blocked by default. Older sites keep the setting their owner chose, so check which you have. We have written about how AI shopping agents decide what to recommend, which is the same question from the other side.
And the Online Safety Act?
If visitors cannot post anything that other visitors see, your site is not the kind of service the Act regulates. A contact form sends a message to you, not to other visitors. If they can only comment on or review your own content, share those comments, react with a like or an emoji, or give a rating, the Act’s “limited functionality” exemption applies (Schedule 1, paragraph 4). A blog with comments on your own posts fits that description.
It is different if visitors can post their own content for other visitors to see, as on a forum, a community board or a marketplace. Then the Act can apply. Ofcom has a free tool, Check if the Online Safety Act applies to you, which asks a few questions and gives an answer.
What to do this month
Compare clicks and impressions in Google Search Console for the last 12 months.
Look for a rise in junk form submissions, the bots your analytics cannot recognise.
Decide on AI crawlers deliberately. If you use Cloudflare, find the setting and make sure it matches what you want.
Run Ofcom’s checker if visitors can post anything that other visitors see.
Frequently asked questions
What does “the death of the open internet” mean?
It is shorthand for at least three different worries. One is governments tightening control through age checks, digital ID and message scanning. Another is a few large platforms and AI companies controlling access to information. The third is AI answers and bots replacing human visits to websites. People using the phrase rarely say which one they mean.
What did Pavel Durov say about the free internet?
In a post on 9 October 2025, the eve of his 41st birthday, Telegram’s founder wrote that “our generation is running out of time to save the free Internet built for us by our fathers”. He listed digital IDs in the UK, online age checks in Australia and mass scanning of private messages in the EU. He also said Germany and the UK were punishing people for what they posted online, and that France was investigating tech leaders.
Does the Online Safety Act apply to my business website?
No, if visitors cannot post anything other visitors see, or can only comment on, review, like, share or rate your own content. Those “limited functionality” services are exempt under Schedule 1, paragraph 4 of the Act. If visitors can post their own content for others to see, such as on a forum or marketplace, use Ofcom’s free checker.
Should I block AI crawlers from my website?
Allowing them lets AI assistants describe and recommend your business. Blocking them keeps your content out of AI training and answers. If your site joined Cloudflare on or after 15 September 2026 and shows ads, training and agent crawlers are already blocked by default.
Is most internet traffic really bots?
Thales’s 2026 Bad Bot Report found automated traffic made up more than 53% of all web traffic in 2025, and bad bots alone 40%. Thales sells bot protection, and the figures cover the traffic it can detect.
Flux Dynamics is a fractional CTO who builds. If your search traffic has changed and you want to know why, tell us about your site and we will give you a straight read of the numbers.