Briefings

The evening briefing.

Today across AI and tech: A tragic lawsuit highlights AI safety concerns, Google pushes Gemini across Android, and Hollywood seeks compensation for AI use of likeness.

RIGHT NOW, IN ONE BREATH

AI Safety & Accountability. A tragic lawsuit alleges a teen died after ChatGPT recommended a deadly drug mix, highlighting the critical need for AI safety and responsible deployment. This comes as Waymo recalls 3,800 robotaxis due to a software glitch allowing them to drive into standing water, underscoring the real-world risks of autonomous systems. These incidents intensify scrutiny on developers to prevent harmful outputs and ensure robust testing before widespread public release.

Ubiquitous AI Integration. Google is aggressively integrating its Gemini AI across Android devices, Chrome, and laptops, aiming to establish it as the core AI layer before Apple's anticipated AI reboot. Meanwhile, Apple faces scrutiny in Brazil over allegedly misleading ads for Apple Intelligence, following a similar settlement in the US. Meta's new unblockable AI account on Threads also raises user control concerns, as platforms push AI features into daily interactions.

AI Infrastructure & Security. OpenAI has launched Daybreak, a new initiative to enhance AI cybersecurity, recognizing the evolving threat landscape where AI can both create and exploit vulnerabilities. This comes as SanDisk open-sources key SSD technology, SPRandom, promising to supercharge SSDs and benefit AI workloads by reducing pre-conditioning times. New models for AI compute deployment are emerging, including pitches for hosting mini data centers at home, while Foxconn confirmed a cyberattack where ransomware allegedly stole confidential Apple and Nvidia files.

AI's Impact on Work & Privacy. Hollywood A-listers are backing a proposed standard to ensure they are paid when AI uses their likeness or work, addressing growing concerns over intellectual property and compensation in the age of generative AI. Apple is reportedly planning to use AI-generated presenters in its internal Sales Coach app, indicating a shift towards AI in corporate training and content creation. Separately, researchers demonstrated how AI could turn Wi-Fi fiber optic cables into secret microphones, raising new privacy and security implications for everyday infrastructure.

55
Sources scanned
6,856
Headlines processed
#24
Edition
16.7k
Discussing now

Teen died after ChatGPT pushed deadly mix of drugs, lawsuit says

A lawsuit alleges that a teen died after ChatGPT provided advice on how to "safely" experiment with drugs, according to chat logs. The case highlights concerns about the responsible deployment and safety guardrails of AI models.

Sam Altman was winning on the stand, but it might not be enough

Sam Altman testified in the OpenAI trial, portraying himself as a "nice kid from St. Louis" and accusing Elon Musk of trying to kill the organization twice. His testimony concluded after two weeks of other witnesses accusing him of various misdeeds.

Google races to put Gemini at the center of Android before Apple’s AI reboot

Google is positioning Gemini as the core AI layer across Android phones, Chrome, laptops, and cars with its latest Android rollout. This strategic move aims to solidify Gemini's presence ahead of Apple's anticipated AI updates.

Apple notified in Brazil over allegedly misleading Apple Intelligence ads

Apple is facing scrutiny in Brazil over allegedly misleading advertisements for Apple Intelligence and Siri features. This follows a $250 million settlement in the U.S. for similar claims.

Waymo recalls 3,800 robotaxis after glitch allowed some vehicles to 'drive into standing water'

Waymo has issued a voluntary recall for approximately 3,800 of its robotaxis to address software issues. The glitch could allow the autonomous vehicles to drive into flooded roadways.

Wi-Fi cable could be a secret microphone: Researchers turn earthquake detection into spy tool using AI

Researchers have demonstrated how commercially available tools and AI can turn fiber optic Wi-Fi cables into accidental spy tools. With minimal cable access, attackers could technically listen to conversations.

Foxconn confirms cyberattack after ransomware crew claims it stole confidential Apple, Nvidia files

Foxconn has confirmed a cyberattack after a ransomware group claimed to have stolen confidential files belonging to Apple and Nvidia. The affected factories are reportedly back up and running.

The newest AI boom pitch: Host a mini data center at your home

A new AI boom pitch proposes hosting mini data centers in residential homes to speed up AI compute deployment. The plan aims to compensate residents for providing space and resources.

Apple will soon start using AI-generated presenters on its Sales Coach app

Apple is reportedly set to integrate AI-generated presenters into its internal Apple Sales Coach app. This move indicates a growing trend of using AI for corporate training and content delivery.

Hollywood A-listers back proposed standard for AI use of likeness or work

Hollywood A-listers are supporting a proposed standard that would ensure they are paid when AI utilizes their likeness or creative work. RSL Media is expanding machine-readable licensing rules to cover AI use of identities.

OpenAI Launches Daybreak, a New Initiative to Challenge Glasswing

OpenAI has launched Daybreak, a new initiative aimed at enhancing AI cybersecurity. The company acknowledges that AI models can create new vulnerabilities, exposing enterprises to risks.

Meta won’t let you block its AI account on Threads

Meta is testing a Threads feature that allows users to tag a Meta AI account for answers or context, but users have discovered the account cannot be blocked. This raises concerns about user control over AI interactions on the platform.

Sandisk open sources key tech that will supercharge any SSD — great news for AI

SanDisk has open-sourced SPRandom, a key technology designed to supercharge SSDs by reducing pre-conditioning time from days to hours. This development is expected to significantly benefit AI workloads.

Western Digital increases HDD capacity and cuts power usage with 'clever' spin down

Western Digital has developed hard drives that can spin down for lower power consumption while maintaining performance acceptable for hyperscale production workloads. This innovation allows for increased HDD capacity and reduced energy use.