Security

Home security company ADT has acquired Origin Wireless, which can use Wi-Fi signals to detect where people and objects are. An appropriately named commenter recognizes there might be some cause for concern.

Lockdown Mode is “not necessary” for most people and “tightly constrains how ChatGPT can interact with external systems to reduce the risk of prompt injection–based data exfiltration,” according to OpenAI.

Yes, Wyze has had its own issues, but this video is pretty funny.

Researchers raised alarms when over 400 malicious skills were uploaded to ClawHub and GitHub in just one week. That prompted an outcry, so OpenClaw partnered with VirusTotal to scan third-party skills. The company acknowledges it’s not a “silver bullet,” but it should provide at least some reassurance to concerned users.

404 Media reports that security researcher Jamieson O’Reilly found a vulnerability that allows humans to control OpenClaw’s AI agents on Moltbook — the network that recently went viral for hosting “discussions” between supposed AI bots.

Two days after Nick Benson asked for donated dashcams in order to document the behavior of federal immigration agents flooding his city, Renee Nicole Good was shot and killed by federal agent Jonathan Ross.

The company says an “unauthorized individual gained access to certain Betterment systems through social engineering” to send the message on Friday. Betterment believes the individual accessed information like “certain names, email addresses, physical addresses, phone numbers, and birthdates,” though so far, its investigation has shown that no passwords were compromised.

Seems a lot of people got password reset requests from Instagram over the last few days, including several Verge staffers and members of their family. The email might look legit. It might even have that little blue checkmark in Gmail. But, it probably came from a scammer. Honestly, it’s best practice to never click links in emails anyway.

In June, Aflac disclosed a data breach involving a “sophisticated cybercrime group” that stole names, social security numbers, contact information, health data, and more from its systems. The insurance provider has now revealed just how many people are affected, adding that it is currently “not aware of any fraudulent use of personal information.”

On Thursday, Trust Wallet announced a “security incident” affecting version 2.68 of its Chrome extension. Binance founder Changpeng Zhao confirmed that Trust Wallet “will cover” the losses and that the team is investigating the hack.

The feature was first piloted in the UK earlier this year, and works by automatically warning users when they launch eligible financial apps while screen sharing during calls with numbers that aren’t in the device’s contact list. The warning forces a 30-second pause period that aims to “break the spell of the scammers’ social engineering,” according to Google.

An exposed dataset from the license plate surveillance company Flock, which is known to work with the US Border Patrol and ICE via local police, showed that some of the AI annotators paid to classify American license plates are located in the Philippines.

The infrastructure buildout will add nearly 1.3 gigawatts of capacity for AI and cost up to $50 billion, the company said. US government customers will have access to both AWS Trainium AI chips and NVIDIA chips, and Amazon said it plans to start building the data centers in 2026.