Is AI Making the Internet Less Open?
Someone still has to read, check and act on everything being produced. When that workload becomes overwhelming, restricting access can start to look attractive.
Faster Writing, More Reviewing
Imagine receiving a polished report with incorrect figures, missing context and references that don’t support its conclusions. Its author may have saved time, but you now have to reconstruct the work.
Researchers at BetterUp and Stanford’s Social Media Lab call this kind of output “workslop”: AI-generated material that looks finished but offers little practical value.
In BetterUp’s September 2025 survey, 40% of respondents believed they had received workslop during the previous month. Respondents reported spending an average of one hour and 51 minutes dealing with each incident. These are self-reported experiences, rather than a measurement of all AI-assisted work.
When an Open Invitation Becomes Unsustainable
The curl software project provides a concrete example. It ended its bug bounty program at the end of January 2026 after struggling with poor-quality security submissions, including AI-generated reports.
Maintainer Daniel Stenberg explained that checking and disproving these reports consumed time and energy. The project removed financial rewards while continuing to accept legitimate vulnerability reports.
That distinction matters: curl changed the incentives around participation rather than closing the project to outside contributors. Read Stenberg’s explanation.
Why Newcomers Could Lose Out
One possible response to overwhelming submissions is to prioritize people with an established reputation. Invitations, referrals and smaller communities can make reviewing contributions more manageable.
The risk is that a talented newcomer has fewer opportunities to demonstrate their ability. Producing something becomes easier, while getting someone to examine it becomes harder.
This is a potential consequence, not evidence that every online community is closing its doors. The outcome depends on how each organization manages quality and access.
AI-Assisted Work Can Still Be Valuable
There is an important follow-up to curl’s story. In April 2026, Stenberg reported that low-quality AI submissions were no longer the same problem. Report quality had improved substantially, and he believed most submissions still involved AI assistance.
His later update shows why treating all AI-generated work as worthless misses the point. Useful findings and careless submissions can arrive through the same tools.
Make Your Contribution Easy to Verify
For writers, developers and anyone sharing AI-assisted work, a practical standard is to reduce the recipient’s workload:
Check factual claims and link to supporting evidence.
Include the context needed to understand your contribution.
Test technical claims before submitting them.
Remove material that adds length without helping the reader.
AI can help more people contribute. Keeping that opportunity open requires contributors to take responsibility for what they send—and communities to leave room for newcomers who do the work carefully.

Comments
Post a Comment