Google researchers say AI is ruining the internet

While Google's product teams are working overtime to shove generative AI into every corner of the web, its own researchers are sounding the alarm on what that actually looks like.

A new, yet-to-be-peer-reviewed paper spotted by 404 Media reveals that Google researchers are warning that generative AI is ruining vast swaths of the internet with synthetic junk and deceptive fakes.

The internet is drowning in cheap fakes

The researchers analyzed existing academic work alongside roughly 200 news reports documenting real-world misuse of generative AI.

Their findings are grim. The vast majority of people misusing the technology are leveraging it to blur the lines between authenticity and deception, primarily through altering human likenesses and fabricating evidence.

Most of these incidents were not harmless pranks. They were launched with clear motives: manipulating public opinion, executing financial scams, or simply chasing cheap ad clicks.

What makes the current wave so dangerous is accessibility. As the researchers pointed out, modern AI systems require minimal technical expertise, meaning virtually anyone can flood social feeds with convincing fabrications that distort socio-political consensus.

The liar's dividend is already here

The researchers also highlighted a deeper psychological toll on everyday internet users.

When malicious and spammy synthetic media floods the web, it does not just fool people. It forces everyone into a state of constant verification overload, driving broad cynicism toward all digital information.

Even worse, it creates a loophole for the powerful. The paper notes cases where high-profile individuals brushed off legitimate, damning evidence by simply claiming it was an AI-generated deepfake.

Interestingly, the paper steered clear of mentioning any of Google's own high-profile AI mishaps, such as recent instances where staff had to manually scrub bizarre search summaries from live results.

Why it matters

This research puts a spotlight on an uncomfortable reality: generative AI is not necessarily being broken by bad actors; it is working exactly as designed.

The technology is exceptionally good at manufacturing plausible illusions at zero marginal cost. When tech giants aggressively distribute these tools to billions of people without adequate guardrails, an explosion of digital pollution is the inevitable outcome.

For regular users, the cost of browsing the web is quietly skyrocketing. You now have to spend mental energy verifying basic screenshots, clips, and headlines, while bad actors exploit the chaos to dodge accountability.

We think this paper exposes a glaring rift inside big tech. While corporate roadmaps push AI into every product to please investors, the companies' own scientists are documenting the slow collapse of online trust that follows.