OPPO's 499-yuan AI pendant records your day from your collar

Share
OPPO's 499-yuan AI pendant records your day from your collar

Two stories with the same shape today: a Chinese phone maker put a price on always-on ambient recording, while the enterprise security vendors spent the day selling controls for the AI their customers already use without asking.

OPPO opened sign-ups for a 9.8-gram AI recorder you clip to your collar, at 499 yuan. The device, called 心力球 ("Heartball"), attaches to a shirt or lapel, records continuously through a rated 24-hour battery, and turns the day into structured notes — events split automatically, each with a title and summary, a short press to mark a highlight, a long press to ask a question, and hooks into office software for meeting notes. OPPO chief product officer Liu Zuohu positioned it at today's Find X10 launch as a "super work partner" that catches what you heard but did not write down, following its first appearance at the company's developer conference on September 17.

The privacy design is where the interesting claim sits. OPPO says the audio is discarded once it has been parsed into text, with the processing run through what it describes as multi-layer encrypted private cloud computing, and the local algorithms handling recognition, memory evolution and cross-source pattern summaries rather than raw recordings. Treat that as a vendor claim until independent testing exists — the device is not shipping yet, and the 499-yuan figure is an experience price for the users OPPO is now recruiting, with general availability promised later this year.

The bigger story is the price and the form factor, not the feature list. Always-on AI recorders have existed in Western markets as expensive novelty hardware with subscription fees; OPPO has put one at roughly the cost of a mid-range pair of earbuds and folded it into a phone ecosystem that millions of buyers already carry. That is a much faster path to ambient AI than a startup pitch deck, and it lands in a market where consent from the people around you has no mechanical solution at all.


Darktrace made its enterprise AI watchdog generally available, and the telemetry behind it is the news. The company said data from about 8,200 customer deployments shows more than 80% of monitored accounts used generative AI services in August, with the average organization touching five different AI providers that month. SECURE AI scans sessions in ChatGPT Enterprise, Claude, Microsoft Copilot and Amazon Bedrock in real time for jailbreak attempts, sensitive-data exposure and indirect prompt injection, ranks each session by risk, and maps AI agents back to the humans behind them.

The examples Darktrace offered from early customers read like a standing indictment of "shadow AI" as a fringe problem. One organization approved a single assistant and found most employees were using unauthorized services instead; another had nearly 90 AI agents built inside a low-code environment with no approval process for creating them; a third had to trigger an immediate legal review after contractors used several AI platforms without oversight. CEO Ed Jennings, who took the role in March, argues the industry's two decades of cataloguing known threats stopped working because the next attack "starts behaving differently from what is normal for the organization."

The control layer is being sold, not solved. Barracuda shipped an AI Data Security product aimed at smaller companies and their managed service providers the same day, covering more than 1,300 generative AI services with default blocking of customer records, credentials and source code, and Teradata expanded its Tera assistant with context and execution tools for governed agents, due in the fourth quarter of 2026. Three vendors restocking the governance aisle in one morning says as much about buyer anxiety as it does about product ambition.

What to watch: whether OPPO ships 心力球 to general retail this year at anything close to 499 yuan, and whether any Western vendor answers at that price.

Would you wear a device that records every conversation within earshot, including the ones you are not part of? Tell us in the comments.

Read more

AI 101 — How do AI video generators work?

AI 101 — How do AI video generators work?

An AI video generator is a model that starts with a grid of random noise and repeatedly refines it into a moving picture that matches your prompt — the same trick image generators use, stretched across time. Nothing is filmed, and nothing is stitched together from clips the model has seen. The pixels in the finished video were invented, frame by frame, all at once. Why it matters right now Kling 4.0, Kuaishou's flagship video model, went fully live on October 15 with up to 30 seconds of vid

Houmo moves compute inside memory — and A20 enters production

Houmo moves compute inside memory — and A20 enters production

Beijing chipmaker Houmo used its partner conference this week to lay out two compute-in-memory routes at once — and to say the first chip built on the newer one is already in production. Houmo's A20, its first chip built on a 3D compute-in-memory architecture, has entered production. The part supports up to 16 chips fully interconnected, claims a 50% improvement in compute utilization over the previous generation, and covers models from 3B to 30B parameters, according to co-founder and product

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kling 4.0 goes live with 30-second native video at 10-bit HDR

Kuaishou's Kling AI has spent two years becoming the video model people actually finish projects with. Its new flagship, Kling 4.0, went fully live on October 15 — and this round is less about prettier frames than about the two things that make a generated shot usable in a real edit: length and control. Kling 4.0 generates up to 30 seconds in a single native pass, outputs 10-bit HDR at 4K, 1080p and 21:9, and takes up to 15 reference assets in one task. That budget splits across ten images, fi

Brain organoids classified Braille at 83% — in a trio, not alone

Brain organoids classified Braille at 83% — in a trio, not alone

Two stories today: human brain tissue on a chip decoded a tactile signal, and the data-centre company behind some of the largest AI contracts in Europe turns out to be six months behind on its flagship site. Human forebrain organoids grown on a microelectrode array classified Braille letters with 61 percent accuracy on their own and 83 percent when the responses of three organoids were combined, in a peer-reviewed study from the University of Bristol. The team took event-based tactile readings