AI Daily Brief
20 events20 source observations
The day in context
The OpenAI board member who fired Sam Altman says we still can’t trust AI CEOs — Helen is a well-known figure in the AI safety and policy community, though she is perhaps best known for having served on the board of OpenAI, when she was one of four board members who voted to fire Sam Altman. That incident, now sometimes referred to as “the blip,” ended with Sam coming back as CEO, Helen resigning, and a new board of directors.
Is AI Making Us Worse at Cybersecurity? - SecurityWeek — Is AI Making Us Worse at Cybersecurity?
Cursor confirms security breach led to malicious Claude Code extension — Cursor confirmed it suffered a security breach that let hackers distribute a malicious Claude Code extension through its marketplace after stealing credentials from AI infrastructure provider Fireworks AI.
Complete briefing
AI models
4The OpenAI board member who fired Sam Altman says we still can’t trust AI CEOs
Published:
Helen is a well-known figure in the AI safety and policy community, though she is perhaps best known for having served on the board of OpenAI, when she was one of four board members who voted to fire Sam Altman. That incident, now sometimes referred to as “the blip,” ended with Sam coming back as CEO, Helen resigning, and a new board of directors.
The OpenAI board member who fired Sam Altman says we still can’t trust AI CEOs
Published:
Helen is a well-known figure in the AI safety and policy community, though she is perhaps best known for having served on the board of OpenAI, when she was one of four board members who voted to fire Sam Altman. That incident, now sometimes referred to as “the blip,” ended with Sam coming back as CEO, Helen resigning, and a new board of directors.
Lawyer upgrades eight-V100 private AI server to run 80% of DeepSeek V4.1 Flash at 4-bit NVFP4 precision — adds 1TB of RAM, an Nvidia RTX 3090, and fixes crashes caused by a bad PSU | Tom's Hardware
Published:
A lawyer with a private AI server built on eight used Nvidia V100 GPUs returned with an update, moving about 80% of its work up from 2-bit to 4-bit NVFP4 precision and bringing in 1TB of RAM and an RTX 3090. The server uses DeepSeek V4.1 Flash with a legal harness, the workflows and checks he wraps around the model, for his law practice.
Lawyer upgrades eight-V100 private AI server to run 80% of DeepSeek V4.1 Flash at 4-bit NVFP4 precision — adds 1TB of RAM, an Nvidia RTX 3090, and fixes crashes caused by a bad PSU | Tom's Hardware
Published:
A lawyer with a private AI server built on eight used Nvidia V100 GPUs returned with an update, moving about 80% of its work up from 2-bit to 4-bit NVFP4 precision and bringing in 1TB of RAM and an RTX 3090. The server uses DeepSeek V4.1 Flash with a legal harness, the workflows and checks he wraps around the model, for his law practice.
Researchers allege DeepSeek, Kimi, and other Chinese AI models were trained using banned Nvidia Blackwell GPUs — Researchers use new GPU 'fingerprinting' technique to identify the GPUs used to train open models | Tom's Hardware
Published:
UnemploymentLabs, an independent research outfit, has come up with strong evidence that those chips were used for training open-weight Chinese AI models, including DeepSeek V4, Kimi K3.1, MiniMax-M3, and Ling-3.1-Flash.
Researchers allege DeepSeek, Kimi, and other Chinese AI models were trained using banned Nvidia Blackwell GPUs — Researchers use new GPU 'fingerprinting' technique to identify the GPUs used to train open models | Tom's Hardware
Published:
UnemploymentLabs, an independent research outfit, has come up with strong evidence that those chips were used for training open-weight Chinese AI models, including DeepSeek V4, Kimi K3.1, MiniMax-M3, and Ling-3.1-Flash.
AI slop replies, ranked by annoyingness
Published:
6) Repeating the post
5) LLM knows more than the account should about a topic
4) Lame semi-joke
3) "X is the part I would keep/everyone misses..."
2) "As a marketer/builder/dentist, I see this in my own..."
1) Questions faking "curiosity"
AI slop replies, ranked by annoyingness
Published:
6) Repeating the post 5) LLM knows more than the account should about a topic 4) Lame semi-joke 3) "X is the part I would keep/everyone misses..." 2) "As a marketer/builder/dentist, I see this in my own..." 1) Questions faking "curiosity"
AI agents
4Is AI Making Us Worse at Cybersecurity? - SecurityWeek
Published:
Is AI Making Us Worse at Cybersecurity?
Is AI Making Us Worse at Cybersecurity? - SecurityWeek
Published:
Is AI Making Us Worse at Cybersecurity?
Adobe AI Strategy Hinges on Being Wherever Creators Work
Published:
Adobe has a message for creatives wondering whether it can keep up: Don’t count it out. With Figma, Canva, and a wave of AI-first upstarts competing for their attention, the company insists that Creative Cloud has always been more than a collection of desktop apps.
Adobe AI Strategy Hinges on Being Wherever Creators Work
Published:
Adobe has a message for creatives wondering whether it can keep up: Don’t count it out. With Figma, Canva, and a wave of AI-first upstarts competing for their attention, the company insists that Creative Cloud has always been more than a collection of desktop apps.
The AI graveyard: a running list of projects and startups that didn't make it | TechCrunch
Published:
Last month, chatbot platform Character.ai filed for bankruptcy just five years after launching. More recently, privacy-focused AI assistant Ollie announced its shutdown a few weeks ago.
The AI graveyard: a running list of projects and startups that didn't make it | TechCrunch
Published:
Last month, chatbot platform Character.ai filed for bankruptcy just five years after launching. More recently, privacy-focused AI assistant Ollie announced its shutdown a few weeks ago.
Do You REALLY Need an AI Agent?
Published:
A framework for deciding when to use AI agents, how many you need, and when they're worth the investment.
Do You REALLY Need an AI Agent?
Published:
A framework for deciding when to use AI agents, how many you need, and when they're worth the investment.
Applied AI
11Cursor confirms security breach led to malicious Claude Code extension
Published:
Cursor confirmed it suffered a security breach that let hackers distribute a malicious Claude Code extension through its marketplace after stealing credentials from AI infrastructure provider Fireworks AI.
Cursor confirms security breach led to malicious Claude Code extension
Published:
Cursor confirmed it suffered a security breach that let hackers distribute a malicious Claude Code extension through its marketplace after stealing credentials from AI infrastructure provider Fireworks AI.
DuckLake 1.1 solves the absurd complexity of data lakes at AI scale - The New Stack
Published:
DuckLake is an open-source table lakehouse format created by Amsterdam-based DuckLabs that acts as a traffic cop for data management by storing table metadata inside a standard SQL database, rather than scattering manifest files across object storage. Unlike traditional table formats, DuckLake treats lakehouse metadata as a relational database problem.
DuckLake 1.1 solves the absurd complexity of data lakes at AI scale - The New Stack
Published:
DuckLake is an open-source table lakehouse format created by Amsterdam-based DuckLabs that acts as a traffic cop for data management by storing table metadata inside a standard SQL database, rather than scattering manifest files across object storage. Unlike traditional table formats, DuckLake treats lakehouse metadata as a relational database problem.
Nearly 20 years after launch, ancient AMD Radeon HD 2000 and 3000 GPUs finally get a Linux fix for corrupted video playback — AI helps squash a decades-old UVD bug | Tom's Hardware
Published:
A newly accepted kernel patch slated for the upcoming Linux 7.5 release finally resolves a long-standing hardware video decoding bug on AMD's ancient R600 architecture, nearly two decades after the cards first hit store shelves. As spotted by Phoronix, the fix targets the Unified Video Decoder (UVD) engine across the RV610, RV620, RV630, RV635, and RV670 silicon families, making up the Radeon HD 2000 and HD 3000 families released starting in 2007.
Nearly 20 years after launch, ancient AMD Radeon HD 2000 and 3000 GPUs finally get a Linux fix for corrupted video playback — AI helps squash a decades-old UVD bug | Tom's Hardware
Published:
A newly accepted kernel patch slated for the upcoming Linux 7.5 release finally resolves a long-standing hardware video decoding bug on AMD's ancient R600 architecture, nearly two decades after the cards first hit store shelves. As spotted by Phoronix, the fix targets the Unified Video Decoder (UVD) engine across the RV610, RV620, RV630, RV635, and RV670 silicon families, making up the Radeon HD 2000 and HD 3000 families released starting in 2007.
ESLint Ends Paid Contributor Program as AI Upends Open Source Collaboration | Socket
Published:
From Rust to Godot and curl, open source projects are tightening contribution rules as AI-generated code strains maintainers and changes how developers collaborate.
ESLint Ends Paid Contributor Program as AI Upends Open Source Collaboration | Socket
Published:
From Rust to Godot and curl, open source projects are tightening contribution rules as AI-generated code strains maintainers and changes how developers collaborate.
Gigabyte unveils its Nvidia Spark-powered AI workstation — AI TOP AERO combines Blackwell architecture GPU and up to 128 GB memory, if you can get one | TechRadar
Published:
Gigabyte has announced the AI TOP AERO, a roughly one-liter desktop built on Nvidia's RTX Spark platform that pairs a 20-core Arm-based Grace CPU with a Blackwell architecture RTX GPU, up to 128GB of unified LPDDR5X memory and up to a 4TB PCIe Gen5 SSD.
Gigabyte unveils its Nvidia Spark-powered AI workstation — AI TOP AERO combines Blackwell architecture GPU and up to 128 GB memory, if you can get one | TechRadar
Published:
Gigabyte has announced the AI TOP AERO, a roughly one-liter desktop built on Nvidia's RTX Spark platform that pairs a 20-core Arm-based Grace CPU with a Blackwell architecture RTX GPU, up to 128GB of unified LPDDR5X memory and up to a 4TB PCIe Gen5 SSD.
The hottest job in AI right now may be counterintelligence as companies hire to protect secrets | Fortune
Published:
The scandal that brought down Jensen Huang as CEO of Nvidia, the AI chip giant he cofounded, has highlighted the growing demand for protection against industrial espionage.
The hottest job in AI right now may be counterintelligence as companies hire to protect secrets | Fortune
Published:
The scandal that brought down Jensen Huang as CEO of Nvidia, the AI chip giant he cofounded, has highlighted the growing demand for protection against industrial espionage.
The downfall of Jensen Huang, AI's $5.7 trillion man | Fortune
Published:
In October, in a converted church in West London, Jensen Huang stepped onto a stage to preach to a packed congregation. The then–Nvidia CEO had a gospel to share with the politicians, tech executives, and journalists, Fortune among them, crowding the room: The AI boom was only getting started, and Nvidia was in town to help bankroll it.
The downfall of Jensen Huang, AI's $5.7 trillion man | Fortune
Published:
In October, in a converted church in West London, Jensen Huang stepped onto a stage to preach to a packed congregation. The then–Nvidia CEO had a gospel to share with the politicians, tech executives, and journalists, Fortune among them, crowding the room: The AI boom was only getting started, and Nvidia was in town to help bankroll it.
Canva claims its new AI image generator is second only to OpenAI — so I put Canva Craft and ChatGPT head-to-head in 5 image tests using their free versions to see which one really comes out on top | TechRadar
Published:
I wanted to find out exactly how Canva’s new AI image generation capabilities compare to ChatGPT’s, so I put the free version of Canva and the free version of ChatGPT head-to-head with five image tests. I gave each tool prompts that were aligned with some of the examples Canva gave in its announcement, like making product photography or blog imagery, and the results are below.
Canva claims its new AI image generator is second only to OpenAI — so I put Canva Craft and ChatGPT head-to-head in 5 image tests using their free versions to see which one really comes out on top | TechRadar
Published:
I wanted to find out exactly how Canva’s new AI image generation capabilities compare to ChatGPT’s, so I put the free version of Canva and the free version of ChatGPT head-to-head with five image tests. I gave each tool prompts that were aligned with some of the examples Canva gave in its announcement, like making product photography or blog imagery, and the results are below.
The Security Jobs Growing Around Enterprise AI Agents
Published:
Two years ago, most company AI use happened in a chat window. Today, AI assistants read shared mailboxes, search SharePoint, open support tickets and, in some teams, write and run code.
The Security Jobs Growing Around Enterprise AI Agents
Published:
Two years ago, most company AI use happened in a chat window. Today, AI assistants read shared mailboxes, search SharePoint, open support tickets and, in some teams, write and run code.
PYMNTS | China’s Singles’ Day Sells Robots, AI Credits and Help at Home
Published:
The real Ning Zetao showed up on a shopping livestream to flex his muscles. So did his artificial intelligence double.
PYMNTS | China’s Singles’ Day Sells Robots, AI Credits and Help at Home
Published:
The real Ning Zetao showed up on a shopping livestream to flex his muscles. So did his artificial intelligence double.
MegaETH Parent Cuts 20% of Staff Amid AI Expansion
Published:
MegaETH co-founder Shuyao Kong said the restructuring affected employees who helped build the project’s blockchain, wallet and ecosystem.
MegaETH Parent Cuts 20% of Staff Amid AI Expansion
Published:
MegaETH co-founder Shuyao Kong said the restructuring affected employees who helped build the project’s blockchain, wallet and ecosystem.
AI research
1Microsoft found a way to turn a helpful AI into a sleeper agent
Published:
Seven neurons controlled a harmless marker, while a separate 60-neuron test predicted a destructive command format.
Microsoft found a way to turn a helpful AI into a sleeper agent
Published:
Seven neurons controlled a harmless marker, while a separate 60-neuron test predicted a destructive command format.