AI news you can use — August 22, 2026
Amazon raised device prices overnight, and the reason is the AI boom
Amazon quietly increased prices across its Echo, Kindle, Fire TV and eero lines, in some cases by more than half. Fortune reported the change and obtained Amazon's explanation: memory and storage components have become significantly more expensive, and the company says it absorbed those increases for as long as it could before adjusting prices. Among the changes catalogued in coverage of the new price list, the entry-level Echo Dot went from $49.99 to $79.99, the 16GB Kindle from $109.99 to $149.99, the Kindle Paperwhite from $159.99 to $199.99, the Fire TV Stick 4K Max from $59.99 to $84.99, and the eero 7 mesh router from $349.99 to $399.99. The underlying cause is a shortage of memory chips, which are being bought in enormous volume for AI data centres; Apple has described the situation as a once-in-a-century event in memory pricing, and PC makers including Dell, HP, Lenovo and Asus have all been adjusting prices for the same reason.
Why it matters: This is the AI build-out arriving in an ordinary shopping basket. Anything with memory or storage inside it — laptops, phones, routers, tablets, security cameras, point-of-sale hardware — is on the same cost curve, and the increases are arriving without announcements. If you have hardware you were planning to replace in the next six months, price it now and buy the items that have not moved yet; if you sell anything with a chip in it, check your supplier costs before you quote a job. Sale events are still likely, so a discount is not evidence that a price has come back down.
LinkedIn says its "seems like AI slop" button is cutting reach
LinkedIn added a button in late July that lets any reader flag a post as machine-written. The company now says readers used it more than a million times in the first two weeks, and that posts its systems classify as AI slop are getting around 40 percent fewer views than they were a few weeks ago. LinkedIn is also adding a line to post analytics telling authors when readers have flagged their post as seeming like AI, and says it has tuned the feed against writing patterns commonly produced by chatbots. The feature followed research suggesting that more than 40 percent of long posts on the platform were fully AI-generated — the highest rate measured on any major platform.
Why it matters: If you or a marketing helper have been posting AI-drafted updates to win reach, the trade has just flipped: the same text now costs you views, and your own analytics will start telling you readers noticed. Using AI to draft is not the problem — publishing something you have not rewritten in your own voice is. Read your last month of posts out loud; anything you would not say to a customer standing in front of you is what the button is being pressed on.
TikTok will pay $400 million to settle a children's privacy case
The Justice Department announced a $400 million settlement with TikTok and ByteDance resolving litigation over the Children's Online Privacy Protection Act, the US law that governs how services handle the personal data of children under 13. Under the terms described in coverage of the announcement, $300 million is paid now and a further $100 million once a court vacates an earlier consent decree covering the app's predecessor. The government filed the underlying complaint in 2024, and the department notes that the company's ownership, management and privacy practices have changed substantially since then. It is among the largest amounts ever recovered in a case brought under that law.
Why it matters: A settlement of this size is a reminder that account settings, not the law, are what actually protect a child on a given afternoon. If there is a child under 13 in your household using any social app, open the app today and check whether they are on a proper child or teen account rather than an adult one, then turn off personalised advertising and location in that account's privacy settings. If you run a business whose website or app is used by children, this is also the law that applies to you — collecting a child's email address for a newsletter without verifiable parental consent is the sort of thing it covers.
Malware turned car dashboard screens into rented internet connections
Security researchers at Kaspersky documented what they describe as the first malware chain built specifically for car head units — the Android touchscreens sold as aftermarket dashboard replacements for navigation, radio and phone mirroring. The attack rode in through the unit's own built-in updater, which was able to install software that had never been on the device, and delivered a three-stage payload. What it did next was not dramatic inside the car: it clicked advertisements in the background and turned the unit into a relay that other people's internet traffic could be routed through, a pattern also reported by BleepingComputer, which notes the supplier says the issue has been resolved. Driving systems were not affected.
Why it matters: A cheap Android screen in a dashboard is a full computer that nobody patches, and the same is true of budget streaming boxes, video doorbells and smart plugs — they are attractive to attackers precisely because they sit on a home or shop network and are never looked at. If you have an aftermarket Android head unit, open its settings and install any pending system update. More generally, put inexpensive connected gadgets on your router's guest network rather than the main one, so a compromised device cannot see the computer you do your banking or invoicing on.
A business school is putting AI stand-ins of its instructors in a paid course
Harvard Business School's Foundry programme runs an eight-week startup bootcamp built around an AI platform, where participants develop a business model, validate a market and build a pitch deck. Reporting on the programme describes what that AI platform includes: video avatars modelled on the school's actual instructors, built by the avatar company HeyGen, which give feedback during practice pitches and mock board meetings. The course costs $699 and still includes live sessions with human instructors; the avatars are positioned as extra practice rather than a replacement, and the school says the format began as a text chatbot and moved to avatars after early participants asked for something more guided.
Why it matters: Paid training is where AI stand-ins will show up first, because the economics are obvious — a recorded expert can hold a thousand simultaneous meetings. That can be genuinely useful for the repetitive part of learning, like rehearsing a pitch until it stops being frightening. It also makes one question worth asking before you or a member of staff pays for any course, coaching package or certification this year: how many hours are live with a person, and how many are with software wearing that person's face? A programme that answers plainly is fine. One that leaves you to find out in week three has told you something about what you actually bought.