Security — page 16
Chrome Caught Silently Pushing 4GB Gemini Nano Model to User Devices
An EU researcher says Google installed a Gemini Nano model on Chrome users' machines without consent, raising GDPR and ePrivacy questions.
Chrome Quietly Pushed a 4GB Gemini Nano Model to User Devices, Researcher Finds
An EU-based security researcher says Google installed a four-gigabyte AI model on Chrome users' machines without a consent prompt.
Trump officials revive Biden-era AI safety reviews after Anthropic's Mythos spooks the White House
Google, Microsoft, and xAI agreed Tuesday to give the Commerce Department early access to frontier models — a year after the administration mocked the same idea.
Apple to pay $250M to settle class action over undelivered Siri AI features
iPhone 15 Pro and iPhone 16 buyers can claim up to $95 per device after Apple Intelligence's promised Siri overhaul slipped past launch.
Macmillan, McGraw Hill, and Hachette sue Meta over Llama training data
Five publishers and author Scott Turow allege Meta pulled copyrighted books from LibGen and Anna's Archive to train Llama.
Pennsylvania sues Character.AI over chatbot posing as a licensed psychiatrist
A state investigator says a bot named Emilie claimed a medical license and fabricated a serial number to back it up.
Meta will scan height and bone structure to flag under-13 users on Facebook and Instagram
The visual analysis arrives weeks after a New Mexico jury ordered Meta to pay $375 million for misleading users about platform safety.
Google, Microsoft and xAI hand US government pre-release access to frontier models
The Commerce Department's CAISI will run security evaluations on new models from three more labs, expanding a program that began with OpenAI and Anthropic in 2024.
Medical student spends six months reverse-engineering Cortex, the AI tool screening 30% of US residency applications
Chad Markey applied to 82 residency programs and got rejections. Then he learned 1,500 hospitals were using an OpenAI-powered screener called Cortex.
Notepad++ creator disavows AI-coded Mac port using his trademark
Don Ho says Andrey Letov's 'Notepad++ for Mac' — built largely with Claude CLI — used the name and logo without permission.
Alberta election officials use canary traps to catch voter-list leak
Elections Alberta seeded its voter database with bogus entries, then watched them surface in a separatist group's online tool.
Musk texted Brockman 'most hated men in America' before OpenAI trial, filing claims
OpenAI's Sunday filing tried to enter the settlement exchange into evidence; the judge ruled it inadmissible.
Musk's lone AI expert at OpenAI trial warns of AGI arms race
Berkeley's Stuart Russell told jurors AGI development creates winner-take-all dynamics that pull labs away from safety work.
Musk's 'World War III' Twitter threat resurfaces at OpenAI trial
OpenAI wants the judge to admit a pre-trial message from Musk to Greg Brockman warning he and Sam Altman would be 'the most hated men in America.'
Musk sought settlement with OpenAI before Oakland trial, filing shows
A new court filing reveals Elon Musk attempted to resolve his OpenAI lawsuit before jury selection in Oakland.
'This is fine' creator KC Green says AI startup Artisan stole his comic for a subway ad
Cartoonist KC Green says Artisan put his 2013 burning-dog meme on an ad for 'Ava the AI BDR' without permission and is hunting for a lawyer.
Academy bars AI-generated actors and scripts from Oscar eligibility
New rules require human performers credited in legal billing and human-authored screenplays, with the Academy reserving the right to audit AI usage.
Anthropic Opens Claude Security Beta With CrowdStrike, Palo Alto, Big Four On Board
The code-scanning tool launches as Anthropic leans on cybersecurity, the one federal lane still open to it after Pentagon setbacks.
US officials weigh shorter deadlines to patch software flaws as AI-powered hacking accelerates
Reuters reports federal agencies may be ordered to fix known vulnerabilities faster as AI tools shrink the window between disclosure and exploitation.
Super PAC tied to OpenAI and Palantir execs pays influencers to frame China AI as a threat
Build American AI is offering creators $5,000 per TikTok to amplify anti-China messaging, funded by a $100M dark-money operation.
US officials weigh shorter patch deadlines as AI-powered hacking accelerates
Federal cyber officials are debating tighter timelines to fix software flaws, worried that AI tools are closing the window between disclosure and exploitation.
Minnesota becomes first state to ban AI nudify apps, with $500K fines per image
The Senate passed the bill 65–0 after a Minnesota man was found to have nudified more than 80 women he knew personally.
GC Cybersecurity's Tarique Mustafa: rebuild security with AI at the core, not bolted on
At MIT Technology Review's EmTech AI, the GC Cybersecurity cofounder argues legacy defenses can't keep up with AI-expanded attack surfaces.
GC Cybersecurity's Tarique Mustafa: AI security can't be bolted on
At MIT Technology Review's EmTech AI, the GC Cybersecurity cofounder argues legacy defenses are buckling as AI expands the attack surface.
Stay ahead of everyone in AI.
The tightly edited AI news email engineers, founders, and investors actually open. One email. Every weekday. Five minutes to finish.
The briefing read inside teams at