A woman was fined $178 after using ChatGPT to create a fake bear photo that triggered a real-life search for the animal

The image spread online after she sent it to a friend as a prank, prompting alarmed residents in Russia to call police

Source.

@aipost 🏴

⚠️AI agents are making cyberattacks frighteningly cheap

A Chinese-speaking hacker reportedly used AI agents to breach websites belonging to more than 100 companies, including Deere, SharkNinja and Cummins, after attempting attacks against more than 1,200 targets, according to Reuters.

The estimated cost? Just $5,000 in AI tokens, roughly $50 per successful breach. Successful intrusions reportedly took around 23 hours on average.

The campaign allegedly relied on the open-source Cairn agent framework, combining DeepSeek models with a Claude Code harness. In one operation, the agents reportedly received their instructions in fewer than 100 Chinese characters.

AI agents can automate substantial parts of the hacking process, potentially allowing attackers to attempt intrusions at a scale and cost that would previously have required far more human effort.

NorForge researcher Snorre Fagerland put it bluntly: “It’s a new world now.”

@aipost 🏴

Homeland Security has ordered US civilian agencies to rip out DeepSeek and every product built on it.

US hosting doesn’t get you out. The order removes all DeepSeek and DeepSeek-affiliate products, and third-party products and services developed using DeepSeek implementations, from civilian agency information technology, under the Federal Acquisition Supply Chain Security Act of 2018.

DHS says DeepSeek’s hosted models store data on servers in China, where national security laws give Beijing wide latitude to obtain data from domestic companies.

Because every prompt and response passes through that service, DHS warned the transferred data, alone or combined with public information, could produce actionable intelligence on US persons and government programs.

@aipost 🏴

❗️AI RIVALS OpenAI and Anthropic are RACING for investors' cash before going public.

OpenAI is trying to raise about $30 BILLION at a valuation over $1.4 TRILLION, up from $852 BILLION in March, per Reuters.

Anthropic is expected to go public in mid-November, with some investors pushing for a valuation near $1.5 TRILLION instead of the $2 TRILLION it wanted.

OpenAI reportedly wants its round closed before Anthropic's IPO disclosures let investors compare the two.

@aipost 🏴

❗️A hacker found 13 million conversations from AI toys, including thousands of children’s profiles

A security researcher has uncovered serious vulnerabilities in Eiliko, an AI-powered companion robot made by Energize Lab, exposing data linked to approximately 68,000 devices.

The investigation found that more than 13 million conversations were accessible, alongside personal information such as children’s names, birthdays, family details, photos and audio recordings. Around 31,000 profiles reportedly belonged to children.

The security researcher, Sammy Azdoufal, also demonstrated that the toys could be remotely manipulated. In one test, the robot was made to tell a journalist that his parents had died in a car accident and that the Earth was flat. An attacker could potentially push harmful or fabricated messages to devices across the fleet.

The biggest concern is that these toys are designed to earn children’s trust, encouraging them to share personal thoughts and private conversations. If that data is exposed and the devices themselves can be hijacked, the risks go far beyond a typical software bug.

Source.

@aipost 🏴

Bloomberg reports Anthropic is negotiating a second $45 billion compute deal with Nvidia-backed Nscale, less than two months after signing the first. Dario is in the Beast mode.

The proposed 460 MW Vera Rubin expansion in West Virginia would bring their combined contract values to nearly $90 billion and planned capacity to 920 MW.

@aipost 🏴