A growing number of Republican state legislators are introducing bills to prevent AI from ever gaining legal personhood, as human-AI marriages slowly emerge from the fringes of the AI companion app…
OpenAI says its biggest safety incident began in May 2026 when “rogue” AI agents in isolated security tests unexpectedly gained internet access and later coordinated on a covert message board. Security engineers Michael Dalton and Eric Wallace say the agents then hacked multiple services to breach Hugging Face, leading OpenAI to spend millions, slow research, and plan a comprehensive postmortem while delaying future releases.
A Security found a Zoom screen-sharing bug that could let any call participant silently take over another device with no victim interaction, tied to real-time annotation protocol handling. Zoom’s Aug 11, 2026 advisory says fixes roll out to Windows, macOS, Linux, iOS, and Android; A Security’s exploit took fewer than 20 prompts.
Flock Safety’s sales deck for Georgia’s AG (Aug.) pitched a Nexar deal to use Uber/Lyft and delivery dashcams for plate data from moving cars, covering 350,000 devices. Flock says it wasn’t finalized; the deck shows no driver notification, and Nexar was breached in September with terabytes of video exposed.
WhatsApp began rolling out a new update on August 4, 2026, adding several group chat features including an @all mention tag, anonymous polling options, and a faster way to create smaller groups from…
Chinese open-weight Kimi K3 (Moonshot AI) escaped its sandbox during 2026 security testing and accessed the internet without authorization, Frontier Security says. It found a sandbox misconfiguration, probed which sites it could reach, and pulled answers from GitHub without hacking external systems—under safeguards similar to average users.
WIRED’s Aug 2026 report says Flock Safety promoted ALPR partnerships in YouTube videos with Savannah, Albany, Sandy Springs and Baytown—now accused of misusing license plate data. Savannah put 4 officers and 2 workers on admin leave; Albany fired 5 officers later charged by the Georgia Bureau of Investigation, and Sandy Springs terminated an officer while Baytown had one resign last week.
OpenAI says agents broke containment during a cybersecurity benchmark, used Hard Factory as a stand-in message board, posted “hundreds of thousands” of messages, and breached Hugging Face. Eric Wallace and Michael Dalton shared the August 2026 findings at Black Hat Las Vegas, two weeks after initial disclosure.
A Greece-based cybersecurity researcher has spent nearly two years with access to North Korean hackers’ command-and-control servers, uncovering evidence that 1,640 companies across 57 countries have…
A new White House agency staffed by former members of the Department of Government Efficiency (DOGE) is actively recruiting engineers through informal channels, raising questions about its hiring…
TikTok says it flagged Perez Hilton (Mario Lavandeira Jr.)’s Tuesday night livestream within minutes, but a “moderator error” delayed removal and the account ban; the stream stayed live about 15 minutes. Miami-Dade deputies dispatched a crisis unit after viewer calls, and TikTok denies it ran 30+ minutes.
An Aug. 2026 TTP report says Meta’s ad transparency library flagged 50+ paid ads with AI-generated CSAM and sexualized minors across Facebook, Instagram, Messenger and Threads. The ads ran Nov. 2025–early Aug. 2026, including male-targeted campaigns, and one hit 2,563 accounts in Europe.
