July 30, 2026
TechCrunch reported that LinkedIn is rolling out a way for users to flag posts that read like AI slop, and that it's also swapping its own AI writing assistant for a proofreading tool instead. Read between the lines and this is LinkedIn admitting its feed has a quality problem it helped create. The platform spent the last few years pushing users toward AI-generated posts, and now it's building a button to clean up the result. That's not hypocrisy exactly, but it is a company recognizing that generated content at scale erodes the thing that made the platform useful in the first place: people wanting to read what other people actually think.
For businesses using LinkedIn as a marketing or sales channel, this matters more than it looks. If a slop-reporting button gains real traction, thin AI-written thought leadership posts are going to get buried or actively flagged by the audience you're trying to reach. The lesson isn't to stop using AI for content -- it's to stop using it as a substitute for having something to say. Teams running content through a marketing workflow should treat AI as a drafting tool, not a publishing tool, and keep a human editing pass non-negotiable before anything goes out under a company's name.
The same week LinkedIn is fighting AI-generated noise, TechCrunch also reported that Google says it fixed more Chrome bugs in June than it had over the prior two years combined, crediting AI tools for the jump -- echoing a pattern experts have flagged at Microsoft too. This is the version of AI adoption that actually works: narrow, verifiable, high-volume tasks where a wrong answer gets caught by a test suite rather than published to a million feeds. Bug detection is exactly the kind of work AI is suited for -- pattern recognition against a known, checkable standard -- and it's a useful reminder that the technology isn't uniformly good or bad. It's good at things with fast feedback loops and much shakier at things that require judgment, taste, or truth.
For any business leaning on AI-generated code or AI-assisted QA, this is a genuinely encouraging data point, though it's worth being honest that finding more bugs isn't the same as having fewer of them in production -- it may just mean the backlog was bigger than anyone realized. Teams evaluating AI-generated code vs. no-code approaches should note that the safest AI use cases right now are still the ones with a clear right answer, not the ones asking a model to write something persuasive from scratch.
TechCrunch also covered the return of Friend, the AI companion wearable, this time with voice capability and a notably higher price tag. A device built around simulated companionship charging more once it can talk back is a small but telling signal: consumer AI hardware is still searching for a business model beyond novelty, and voice is apparently what founders think justifies a premium. It's a fair bet, but it's also a reminder that plenty of consumer-facing AI products are testing willingness-to-pay in real time, with mixed results so far. Business buyers should read this as a caution against assuming every AI feature justifies its markup -- the same scrutiny applies whether you're buying a $99 gadget or evaluating a five-figure software contract, and it's worth checking what a buy vs. build comparison actually shows before paying for AI branding alone.
Which of these trends worries you more as a business buyer: AI content flooding the channels you use to reach customers, or AI hardware charging premium prices before it's proven what people will actually pay for?
Sources