One team says ten days to break the internet's core encryption. Another says nine minutes for Bitcoin. Neither required nearly the resources researchers expected a year ago.
The AI system that selected the first 1,000 strike targets in Iran has a reported accuracy of 60%. No independent body evaluated whether it was reliable enough to decide who lives and who dies.
Big Tech locked in $635 billion for AI infrastructure this year. Then the Strait of Hormuz closed, 11 million barrels of daily oil supply vanished, and the IEA called it the largest energy disruption in history. Nobody has revised the spreadsheets.
The Pentagon spent 16 years and $8 billion building GPS control software. It still can't operate the satellites. The contract option expires today.
Nearly three million OkCupid users had their photos handed to an unrelated company with no restrictions on how the data could be used. The company then spent more than ten years covering it up.
Developers using AI coding tools report a new kind of burnout: not from writing code, but from babysitting the machines that write it for them. One Copilot user found an ad inserted into his pull request.
US District Judge Jane Boyle dismissed Elon Musk's lawsuit with prejudice, ruling that advertisers who walked away from X didn't violate antitrust law — they just didn't want to buy ads there.
Clearview AI flagged a Tennessee grandmother as a bank fraud suspect in North Dakota — a state she'd never visited. She spent five months in jail. Nobody checked her bank records.
An 85-second deepfake shows a Texas Senate candidate saying things he never said. The disclosure is text so small you'd miss it scrolling past — and no federal law exists to stop the next one.
Eleven co-founders. Zero remaining. The last one left the same week Musk admitted the company wasn't 'built right the first time around.'
The US embassy in Mexico published AI-generated musicians performing a traditional corrido that tells migrants to self-deport. Mexico's response was not gratitude.
Indonesia became the first non-Western country to ban under-16s from YouTube, TikTok, Instagram, and five other platforms. Nobody has explained how it will verify anyone's age.
A Texas company promised millions of crime tipsters their identities were safe. Hackers just published 93GB of data — including names, addresses, and Social Security numbers — stored in plain text.
The vote was 40 to 2. But the new policy warns that some human editors write like machines — and catching AI text from style alone is explicitly discouraged.
A threat actor claims to have stolen over 350GB of data from the EU executive body's Amazon cloud environment — including employee databases and email server access. The Commission confirms the breach but says internal systems remain untouched.
The head of America's top law enforcement agency was using personal Gmail. Iranian hackers found it. Now they're publishing what they found — and it's part of a much bigger digital offensive.
A misconfigured CMS exposed details of Anthropic's 'most powerful model ever' — one the company warns could supercharge cyberattacks. Cybersecurity stocks tanked. The irony writes itself.
The AI czar who tried to ban every state AI law in America has been moved to an advisory council. Silicon Valley just lost its direct line to the Oval Office.
The Pentagon designated an American AI company as a 'supply chain risk'—a label usually reserved for foreign adversaries. A federal judge just said that looks a lot like punishment for protected speech.
Apple will let users route Siri queries to Claude, Gemini, and other rival AI services—breaking OpenAI's exclusive grip. The shift marks the biggest change to Siri's strategy since its 2011 debut.
lop News
Jonny V. Nuovo